See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/357813905

Is Diffusion Tensor Imaging-Guided Radiotherapy the New State-of-the-Art? A

Review of the Current Literature and Technical Insights

Article in Applied Sciences · January 2022

DOI: 10.3390/app12020816

CITATION READS

1 146

6 authors, including:

Jordan Colman Spyros Manolopoulos

Royal Surrey County NHS Foundation Trust NCCC (Carlisle)
19 PUBLICATIONS 180 CITATIONS 61 PUBLICATIONS 550 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Spyros Manolopoulos on 23 February 2022.

The user has requested enhancement of the downloaded file.

Review

Is Diffusion Tensor Imaging‐Guided Radiotherapy the New

State‐of‐the‐Art? A Review of the Current Literature
and Technical Insights
Jordan Colman 1, Laura Mancini 2,3, Spyros Manolopoulos 4, Meetakshi Gupta 5, Michael Kosmin 6,7
and Sotirios Bisdas 2,3,7,*

1 UCL Queen Square Institute of Neurology, Faculty of Brain Sciences, University College London,
London WC1N 3BG, UK; [email protected]
2 The Lysholm Department of Neuroradiology, National Hospital for Neurology and Neurosurgery,

Queen Square, Holborn, London WC1N 3BG, UK; [email protected]

3 Neuroradiological Academic Unit, Department Brain Repair and Rehabilitation, UCL Queen Square

Institute of Neurology, London WC1N 3BG, UK

4 Department of Medical Physics & Biomedical Engineering, University College London, Malet Place

Engineering Building, London WC1E 6BT, UK; [email protected]

5 Department of Radiotherapy, University College London Hospitals NHS Foundation Trust,

London NW1 2PG, UK; [email protected]

6 Department of Clinical Oncology, University College London Hospitals NHS Foundation Trust,

London NW1 2PG, UK; [email protected]

7 National Institute for Health Research University College London Hospitals (UCLH) Biomedical Research

Centre, London W1T 7HA, UK

* Correspondence: [email protected]
Citation: Colman, J.; Mancini, L.;
Manolopoulos, S.; Gupta, M.;
Abstract: Despite the increasing precision of radiotherapy delivery, it is still frequently associated
Kosmin, M.; Bisdas, S. Is Diffusion
with neurological complications. This is in part due to damage to eloquent white matter (WM)
Tensor Imaging Guided
tracts, which is made more likely by the fact they cannot be visualised on standard structural imag‐
Radiotherapy the New
State‐of‐the‐Art? A Review of the
ing. WM is additionally more vulnerable than grey matter to radiation damage. Primary brain ma‐
Current Literature and Technical lignancies also are known to spread along the WM. Diffusion tensor imaging (DTI) is the only in
Insights. Appl. Sci. 2022, 12, 816. vivo method of delineating WM tracts. DTI is an imaging technique that models the direction of
https://doi.org/10.3390/app12020816 diffusion and therefore can infer the orientation of WM fibres. This review article evaluates the cur‐
rent evidence for using DTI to guide intracranial radiotherapy and whether it constitutes a new
Academic Editor: state‐of‐the‐art technique. We provide a basic overview of DTI and its known applications in radi‐
Chang Ming Charlie Ma otherapy, which include using tractography to reduce the radiation dose to eloquent WM tracts and
using DTI to detect or predict tumoural spread. We evaluate the evidence for DTI‐guided radio‐
Received: 16 December 2021
therapy in gliomas, metastatic disease, and benign conditions, finding that the strongest evidence
Accepted: 11 January 2022
is for its use in arteriovenous malformations. However, the evidence is weak in other conditions
Published: 13 January 2022
due to a lack of case‐controlled trials.

Publisher’s Note: MDPI stays neu‐

Keywords: stereotactic radiotherapy; image‐guided radiotherapy; diffusion tensor imaging
tral with regard to jurisdictional
claims in published maps and institu‐
tional affiliations.

1. Introduction
1.1. Intracranial Radiotherapy
Copyright: © 2022 by the authors. Li‐
Radiotherapy continues to advance to be able to target areas with higher conformal‐
censee MDPI, Basel, Switzerland.
ity. The introduction of stereotactic radiotherapy (SRT), sometimes referred to as stereo‐
This article is an open access article
tactic radiosurgery, has meant that radiation can be directed with increasing precision.
distributed under the terms and con‐
This results in better outcomes by enabling a higher dose to the area of pathology while
ditions of the Creative Commons At‐
tribution (CC BY) license (https://cre‐
reducing the radiation dose to healthy brain structures. Popular implementations include
ativecommons.org/licenses/by/4.0/).
the Gamma Knife and Cyberknife platforms [1].

Appl. Sci. 2022, 12, 816. https://doi.org/10.3390/app12020816 www.mdpi.com/journal/applsci

Appl. Sci. 2022, 12, 816 2 of 16

Intracranial radiotherapy requires planning the target area, which is typically done
using MRI to identify the target region and eloquent brain structures that need to be
avoided. This is usually performed with structural MRI sequences such as T1w, T2w, and
FLAIR. This is achieved by manual segmentation of the pathology, with eloquent areas
additionally segmented as organs at risk (OAR). Examples of OAR that are currently ad‐
vised to be included are the brain stem, optic chiasm, cochlea, and others, with each OAR
having an advised maximum radiation dose measured in Gray (Gy) to avoid significant
late dysfunction and neurological toxicities [2]. Then, a CT with a frame is required for
dose treatment planning.
With the advent of intensity‐modulated radiation therapy (IMRT), the dose to patho‐
logical structures can be conformed to in 3D with excellent conformality. IMRT splits the
beams into many smaller beamlets, enabling radiation to be directed into complex shapes
and have a non‐uniform radiation intensity [3]. IMRT additionally utilises computerised
inverse planning where the radiation dose to the target region and dose limits to healthy
brain and OAR are specified and optimised in advance. Both Cyber and Gamma Knife
SRT delivery platforms include treatment planning software that enables integrated SRT
planning allow for a combination of frameless MRI and the planning CT [4].
When targeting primary intracranial malignancies, the tumour visible on imaging is
referred to as the gross tumour volume (GTV); this is then expanded by a standard margin
(dependent on the tumour type) to incorporate any peritumoral infiltration that is unseen
on imaging and is referred to as the clinical target volume (CTV). Therefore, the expansion
of the CTV is important to target these migrating cells to reduce the chance of recurrence.
Then, the CTV is typically expanded further by a few mm to allow for patient set‐up errors
that can occur despite immobilisation and be used as the final planning target volume
(PTV) [5]. An example of the differing target volumes is shown in Figure 1. These expan‐
sion steps are typically done isotropically unless there are barriers to tumour spread e.g.,
skull or dura when delineating the CTV. The European Organization for Research and
Treatment of Cancer (EORTC) current guidance for SRT in glioblastoma treatment is to
define the GTV as the T1w contrast‐enhancing tumour, or resection cavity, and then, the
CTV is defined as the GTV expanded isotropically by 2 cm. Then, 3–5 mm is additionally
added to define the PTV [2]. The CTV is set at a 2 cm isotropic expansion, as roughly 80%
or more recurrences are within 2 cm of the enhancing tumour core.

Figure 1. Diagram displaying an example of the gross tumour volume (GTV) with the isotropically
expanded clinical target volume (CTV) and the planned target volume (PTV) on a temporal lobe
glioma. Addition of the brain stem as an organ at risk (OAR) on the right and how the PTV could
be altered using inverse planning to reduce radiation dose to the brain stem.
Appl. Sci. 2022, 12, 816 3 of 16

Radiotherapy to the brain can lead to many neurological complications due to the
damage of healthy brain structures and is usually dependent on dose, fraction size, vol‐
ume of irradiated tissue, and concurrent chemotherapy use [6]. Complications include
vascular abnormalities such as radiation necrosis and ischemia but additionally include
progressive neurocognitive decline. This can lead to many long‐term side effects in addi‐
tion to short‐term complications [7]. WM is additionally more vulnerable than grey mat‐
ter, and some of the cognitive and functional negative effects are thought to be due to WM
tract damage [8,9]. Individual WM tracts are not visible on structural imaging, and there‐
fore, eloquent tracts may be unknowingly given high doses of radiation when in close
proximity to the target region.
Gliomas as mentioned are infiltrative, and target volume expansion is included to
allow for this while weighing up the risks of radiation. However, glioma cells are thought
to spread preferentially along white matter (WM) tracts [10]. Therefore, it has been sug‐
gested that the modelling of WM properties such as their orientation may mean the CTV
can be expanded anisotropically in order to reduce the overall radiation does while still
covering the same proportion of potential recurrence sites [11].
The only in vivo method of delineating WM tracts is diffusion tensor imaging (DTI).
DTI can be used to infer the isotropic and anisotropic properties of white matter and its
orientation. This can be extended to produce virtually dissected WM tracts, the technique
being called tractography. The incorporation of information on WM orientation and WM
tracts appears to have potential for improving PTV delineation in SRT.

1.2. Diffusion Tensor Imaging

In diffusion‐weighted MRI, a gradient is applied in a given direction; if protons dif‐
fuse along the gradient direction, dephasing will be accelerated, attenuating the MR sig‐
nal. Areas with more free diffusion of water will have a reduced MR signal due to the
greater movement of the protons; e.g., CSF will appear dark compared to grey matter as
CSF contains minimal obstructive microstructure. Some tissues such as white matter will
have preferential diffusion in a particular direction, which is known as anisotropic diffu‐
sion. This is due to the WM axons restricting the diffusion of water perpendicular to them,
but diffusion parallel to their orientation being unrestricted. The imaging gradient direc‐
tion can be changed, and the DWI signal would be expected to be higher if perpendicular
to a WM tract and lower if parallel to it [12].
If DWI images are obtained with at least six gradient directions, the orientation or
overall direction of diffusion in the voxel can be calculated and is known as diffusion ten‐
sor imaging (DTI). If the voxel is in WM, then the diffusion direction can be assumed to
be the orientation of the WM tract. The amount to which that diffusion is preferential to
one direction can be quantified and is known as fraction anisotropy (FA), where 0 is com‐
pletely equal diffusion in all directions and 1 is completely uniform diffusion in only one
direction. The overall diffusion of a voxel is known as mean diffusivity (MD). Another set
of similar metrics to MD and FA are p and q maps, which refer to the isotropic and ani‐
sotropy components of the voxel, respectively [13]. A diagram displaying the effects of
tissue microstructure on diffusion measurements is shown in Figure 2.
Appl. Sci. 2022, 12, 816 4 of 16

Figure 2. Diagram displaying tensors from different microstructures and how this would affect the
isotropic tensor derivatives (p and MD) and the anisotropic components (q and FA).

As displayed in Figure 2, DTI can capture microstructural changes of white matter,

as damage leads to reduced FA and increased MD values due to a reduction in axon den‐
sity and reduced myelination. This has been shown to be useful in detecting tumoural
invasion, which can be observed as reduced FA or q values and has additionally been
shown to be potentially useful in differentiating glioma reoccurrence from radiation ne‐
crosis after radiotherapy [14]. MD or p values will also increase with tumour invasion and
have been suggested to be useful in defining glioma margins [15]. Other DTI measures
have additionally been suggested to have potential in the diagnosis of intracranial malig‐
nancies, with ADC being shown to be useful in differentiating brain stem gliomas from
medulloblastoma [16,17].
The information from DTI can be used to model whole WM tracts and is known as
tractography. This is where each voxel is followed from a seed point and walks in the
orientation or tract direction of each voxel; this is repeated for a set number of steps, and
a tract is formed. Then, this is done for each voxel in a seed region. When a single tract is
taken from each seed voxel, this is known as deterministic tractography. An issue with
this method is that a false tract may be followed by accident. Another method known as
probabilistic tractography repeats the tract formation multiple times, and a probability
distribution of potential tracts is taken. Probabilistic tractography is usually more accu‐
rate, as it incorporates an amount of error, and tracts with a lower probability, that are
likely false positives, can be filtered out [18]. Figure 3 displays an example of tractography
forming a tract.
Selecting a seed point to form a desired tract is usually done based on structural anat‐
omy likely to only contain the tract. An example is using the precentral gyrus as a seed
point for the corticospinal tracts, as is contains the primary motor cortex. Way points can
additionally be used, which only includes tracts that pass through them from the seed
point. An example for the corticospinal tract could be the cerebral peduncles.
A limitation of tractography is that when the voxel contains multiple WM tracts with
different orientations, then the mean direction of the two tracts may be taken or a smaller
tract can be lost, as only the overall diffusion direction is taken. This is a particular issue
when two tracts cross or “kiss” in the same voxel and may lead to false positive and false
Appl. Sci. 2022, 12, 816 5 of 16

negative tracts. There are several techniques that attempt to model multiple WM tracts in
one voxel, all of which require a larger number of directions [18]. The main drawback of
these methods is that image acquisition and processing will take longer, but the sequence
acquisitions and the processing software are continuously improved, and with more ad‐
vanced hardware and software, these advanced acquisitions and processing have become
clinically feasible.

Figure 3. Diagram explaining the tractography process showing an example of how the voxel ten‐
sors (left) can be tracked from a seed point to form a tract (middle) and performed multiple times
to produce a virtually dissected white matter tract (right).

2. Review Aims and Objectives

DTI can provide additional information on WM anatomy and therefore has the po‐
tential to improve radiotherapy accuracy. We decided to undertake a literature review on
the subject of DTI‐guided intracranial radiotherapy and whether its use can be considered
a new state‐of‐the‐art for clinical practice. The MEDLINE database via PubMed was
searched for relevant articles along with references from a recent systematic review article
[19]. A systematic review was not performed, and instead, a general overview of the key
literature can be undertaken. We provide an overview of the current uses of DTI for guid‐
ing radiotherapy, which includes the virtual dissection of WM tracts and modelling tu‐
moural invasion. Then, we go onto discuss the evidence for application of DTI guidance
in radiotherapy treatment of gliomas, brain metastasis, and in benign conditions. We ad‐
ditionally provide our own suggestions for future work that should be undertaken based
on the findings in the literature and outline the limitations for the use of DTI in radiother‐
apy.

3. Review Results
Table 1 provides an overview of the key articles discussed in this review and the
subsections to which they are relevant. The articles are discussed in detail below.
Appl. Sci. 2022, 12, 816 6 of 16

Table 1. This table outlines the key articles on DTI‐guided intracranial radiotherapy discussed in
the review broken down into subtopics. AVM = arteriovenous malformation, CST = corticospinal
tract, CM = cerebral metastasis, Menig = Meningioma, OAR = organ at risk, HGG = high‐grade gli‐
oma, VS = vestibular schwannoma.
Number of
Reference Type of Study Summary Main Outcomes
Subjects
Virtual dissection of Eloquent WM Tracts
Tractography used to dissect multiple eloquent
Significant reduction in maximum and mean
Altabella et al. Theoretical plan‐ 19 (all with tracts bilaterally (CST, SLF, IFOF, UNF). Tracts in‐
radiation dose to the tracts, particularly to
2018 [9] ning study HGG) cluded as OAR during inverse planning with
contralateral tracts.
TomoTherapy.
DTI tractography performed using ‘StealthViz’
20 (5 VS, 5
Gavin and Sabin Clinical feasibil‐ used on user selected tracts which were included Methodologies outlined and shown to be
AVM, 9 CM, 1
2016 [20] ity study as OAR into GammaPlan prior to GKRS on a se‐ clinically feasible.
Menig)
ries of cases.
Tractography used to include pyramidal tract as Maximum radiation dose to the pyramidal
Kawasaki et al. Theoretical plan‐ 23 (20 CM, 3
OAR in GKRS planning and plans with and with‐ tracts significantly reduced when tractog‐
2017 [21] ning study AVM)
out tractography compared. raphy used in planning.
52 (all with Integration of tractography into radiotherapy
Significantly less motor complication in pa‐
Koga et al. 2012 Retrospective co‐ AVM, 24 trac‐ planning for AVM at a single centre with 28 con‐
tients with tractography integration. No sig‐
[22] hort study tography, 28 trol cases being prior and 24 test cases after this.
nificant difference in treatment success.
control) Patients followed up for a minimum of 3 years.
Significant reduction of the maximum radia‐
Sun et al. 2017 Theoretical plan‐ 16 (6 AVM, 8 Integration of fMRI to with tractography to model
tion dose to the included cortical areas and
[23] ning study CM, 2 Menig) the corticospinal tracts and ‘sensory pathway’.
tracts.
DTI modelled tumoural invasion
Use of DTI p and q maps and tractography to de‐
The DTI‐defined CTV was significantly
Berberat et al. Theoretical plan‐ 13 (all with glio‐ fine the CTV. Comparison to CTV defined by oe‐
smaller than the T2w MRI‐defined CTV and
2014 [24] ning study blastoma) dema on T2w MRI or isotropic expansion of the
still included sites of tumour recurrence.
GTV.
Produces model using DTI to model tumour
Hathout et al. Theoretical mod‐ DTI‐based 3D mathematical model of glioblas‐
NA infiltration shown to predict glioma growth
2016 [25] elling study toma growth.
in example cases.
Study that develops open‐source DTI based fibre‐
The software is shown to qualitatively cap‐
Jordan et al. Theoretical mod‐ tracking software that produces anisotropic CTV
NA ture areas of recurrence well in a few exam‐
2019 [26] elling study to better capture likely areas of tumour infiltra‐
ple cases.
tion.
Use of deep learning to correct FA maps for free Area under the curve for recurrence predic‐
Metz et al. 2020 Retrospective ob‐ 35 (all with glio‐
water and used to predict areas of glioma recur‐ tion of 0.77 using FA and 0.9 using free wa‐
[27] servational study blastoma)
rence on follow‐up imaging. ter‐corrected FA values.
Free water corrected FA map‐based GTV
Use of deep‐learning‐based free water‐corrected were significantly smaller than traditionally
Peeken et al. Theoretical plan‐ 33 (all with glio‐
FA maps to define the GTV and compared to tra‐ defined GTV but still include the recurrence
2019 [28] ning study blastoma)
ditionally defined GTV. area of all but one of the 14 subjects with re‐
currence.
Semi‐automated DTI‐defined tumour volume.
GTV manually defined using q map, which was
Mean Dice coefficient of 74% between man‐
Rahmat et al. Theoretical plan‐ 50 (all with glio‐ automatically expanded using an expansion
ual and semi‐automated method over all 50
2020 [15] ning study blastoma) model on the p map to model tumour infiltration.
patients.
Comparison made to manual segmentation of p
and q maps.
Applications in Gliomas
Tractography used to dissect multiple eloquent
tracts bilaterally (CST, SLF, IFOF, UNF). Tracts in‐ Significant reduction in maximum and mean
Altabella et al. Theoretical plan‐ 19 (all with
cluded as OAR during inverse planning with radiation dose to the tracts, particularly to
2018 [9] ning study HGG)
TomoTherapy. Treatment plans with and without contralateral tracts.
tractography were compared.
Integration of tractography into radiotherapy
Igaki et al. 2014 Clinical feasibil‐ Integration of tractography shown to be fea‐
NA planning in glioblastoma by including the cortico‐
[29] ity study sible and reduce radiation does in two cases.
spinal tracts as OAR in planning.
Appl. Sci. 2022, 12, 816 7 of 16

Use of deep learning to correct FA maps for free Area under the curve for recurrence predic‐
Metz et al. 2020 Retrospective ob‐ 35 (all with glio‐
water and used to predict areas of glioma recur‐ tion of 0.77 using FA and 0.9 using free wa‐
[27] servational study blastoma)
rence on follow up imaging. ter‐corrected FA values.
Finds only three articles evaluating DTI in ra‐
Yahya and Systematic re‐ Systematic review of literature on DTI in intracra‐ diotherapy that includes gliomas, none of
NA
Manan 2019 [19] view nial radiotherapy. which were case control trials or used pro‐
spective integration.
Applications in Brain Metastases
25 (10 AVM, 3 fMRI and tractography of eloquent structures in Found an average reduction in radiation dose
Conti et al. 2013 Theoretical plan‐
CM, 12 ‘brain close proximity to the lesions. Comparison of radi‐by 17% to eloquent regions when fMRI and
[30] ning study
tumours’ otherapy plans with and without integration. tractography integrated.
20 (5 VS, 5 Applied to nine cases of cerebral metastasis
Gavin and Sabin Clinical feasibil‐
AVM, 9 CM, 1 DTI tractography used to guide GKRS. successfully; however, there was no compari‐
2016 [20] ity study
Menig) son to a control group.
Twenty out of 23 subjects had CM. The maxi‐
Tractography used to include pyramidal tract as
Kawasaki et al. Theoretical plan‐ 23 (20 CM, 3 mum radiation dose to the pyramidal tracts
OAR in GKRS planning and plans with and with‐
2017 [21] ning study AVM) significantly reduced when tractography was
out tractography were compared.
used in planning.
Integration of fMRI to select tractography seed
point to model the corticospinal tracts and ‘sen‐
Significant reduction of the maximum radia‐
Sun et al. 2017 Theoretical plan‐ 16 (6 AVM, 8 sory pathway’. Functional regions and tracts in‐
tion dose to the included cortical areas and
[23] ning study CM, 2 Menig) cluded as OAR during planning for CyberKnife
tracts.
radiosurgery. Treatment plans with and without
integration were compared.
Finds five articles evaluating DTI in radio‐
Yahya and Systematic re‐ Systematic review of literature on DTI in intracra‐
NA therapy that includes CM, one of which uses
Manan 2019 [19] view nial radiotherapy.
prospective integration.
Applications in Benign Conditions
Applied to five cases of vestibular schwan‐
20 (5 VS, 5
Gavin and Sabin Clinical feasibil‐ noma, five cases of AVM, and one case of
AVM, 9 CM, 1 DTI tractography used to guide GKRS.
2016 [20] ity study meningioma successfully; however, there
Menig)
was no comparison to a control group.
Three out of 23 subjects had AVM. The maxi‐
Tractography used to include pyramidal tract as mum radiation dose to the pyramidal tracts
Kawasaki et al. Theoretical plan‐ 23 (20 CM, 3
OAR in GKRS planning and plans with and with‐ significantly reduced when tractography was
2017 [21] ning study AVM)
out tractography were compared. used in planning. However, the plan not
used for treatment.
Integration of tractography into radiotherapy
52 (all with
planning for AVM at a single centre in 2004 with Significantly less motor complication in pa‐
Koga et al. 2012 Retrospective co‐ AVM, 24 trac‐
28 control cases being prior to this and 24 test tients with tractography integration. No sig‐
[22] hort study tography, 28
cases after this. Patients followed up for minimum nificant difference in treatment success.
control)
of 3 years.
155 (all with Routine integration of tractography in radiother‐ Tractography used in 71 out of 155 radiother‐
Koga et al. 2012 Prospective inte‐ AVM, 71 of apy of AVMs at a single centre. Selected tracts apy cases during the study period with 60%
[31] gration which had trac‐ user selected if suspected to be close to the pathol‐ of cases using tractography finding the criti‐
tography) ogy. cal tracts within 5 mm of the lesion.
Integration of fMRI to select tractography seed
point to model the corticospinal tracts and ‘sen‐
Significant reduction of the maximum radia‐
Sun et al. 2017 Theoretical plan‐ 16 (6 AVM, 8 sory pathway’. Functional regions and tracts in‐
tion dose to the included cortical areas and
[23] ning study CM, 2 Menig) cluded as OAR during planning for CyberKnife
tracts.
radiosurgery. Treatment plans with and without
integration were compared.
Finds 13 articles evaluating DTI in radiother‐
Yahya and Systematic re‐ Systematic review of literature on DTI in intracra‐
NA apy that include AVM, many of which use
Manan 2019 [19] view nial radiotherapy.
prospective integration.
Appl. Sci. 2022, 12, 816 8 of 16

3.1. Virtual Dissection of Eloquent WM Tracts

The benefits of knowing the position of eloquent WM tracts are immediately clear, as
if known, their radiation exposure can be minimised, and more of their function pre‐
served. This concept has already been used in stereotactic neurosurgery to avoid im‐
portant WM tracts [18]. This can be applied to radiotherapy to include eloquent tracts as
OAR and minimise their radiation dose. However, the possible tracts passing near the
area of pathology may not be obvious, and disease processes such as tumours may alter
or obscure the normal anatomy. Therefore, the application of tractography requires a
highly skilled operator and is potentially subjective. An example of altering a CTV by
including a tract as an OAR is shown in Figure 4.
A study used deterministic tractography in Gamma Knife radiosurgery to model sus‐
pected tracts at risk, such as the optic radiations and arcuate fasciculus, including them as
OAR in target volume planning, showing that this is possible. However, one limitation is
that the study does not make any comparison to a control group to show that the tracts
dose is reduced [20]. Additionally, it remains unclear whether reducing radiation dose to
this part of the standard clinical target volume leads to higher rates of local recurrence or
not.
Several examples in the literature focus on methods of minimising radiation to the
pyramidal tract/corticospinal tract (CST), due to it being one of the most salient WM path‐
ways and being easily identified with the more basic DTI processing. A study compared
the predicted radiation does to the pyramidal tract with and without the use of including
the tracts. The study found there was a significantly reduced dose to the pyramidal tract
without dose reductions in the area of pathology [21]. The study used the cerebral pedun‐
cles, posterior limb of the internal capsule, and primary motor cortex selected on T1w MRI
as the seed points for tractography. Another retrospective study that used a non‐random‐
ised control method with 52 subjects in total found that modelling of the corticospinal
tracts reduced the risk or motor complications in patients treated with SRT, although not
all non‐control subjects had the CST dissected, and it was operator dependent [22].
A limitation of using structural anatomy for selecting seed points is that functional
areas are not always anatomically consistent. A paper attempted to overcome this issue
by using fMRI to define the seed region of the pyramidal tract. They compared radiother‐
apy plans with and without the inclusion of the CST as an OAR and found that their in‐
clusion reduce the average radiation dose by 22.7% [23]. However, a limitation of this
study is that fMRI and anatomically defined pyramidal tracts were not compared. Addi‐
tionally, the two techniques were not compared when used clinically, so the clinical out‐
comes of the two techniques could not be compared.
One paper attempted to model multiple tracts, including those that are important for
cognition. The study used tractography to segment several tracts bilaterally including the
superior longitudinal fasciculus (SLF), the arcuate fasciculus, the inferior fronto‐occipital
fasciculus (IFOF), and the uncinate fasciculus. The study compared treatment plans with
and without including the dissected tracts as OAR and found that radiations doses were
significantly reduced if included [9].
Appl. Sci. 2022, 12, 816 9 of 16

Figure 4. Diagram displaying (A) Example without tractography dissected corticospinal tract (CST)
included as an OAR, and (B) CST included as an OAR during dose planning, resulting in a reduced
radiation dose.

3.2. DTI Modelled Tumoral Invasion

Gliomas typically spread along the white matter tracts with tumour cells migrating
with the orientation of the axons [32,33]. Typically, the tumoral core in high‐grade tu‐
mours is defined as the contrast‐enhancing tumour often with necrotic tissue within it [2].
However, glioma cells are found outside this core region due to the tumoral spread. As
glioma cells migrate, it is thought to cause a loss of axonal integrity, leading to increased
diffusivity and the T2 signal, and it is difficult to differentiate from vasogenic edema
[34,35]. It has been suggested that diffusion imaging can improve the identification of non‐
enhancing tumour infiltration. For example, Pavlisa et al. 2009 find that the ADC is lower
in the peritumour tissue of infiltrative tumours compared to non‐infiltrative tumours [36].
Another study additionally found FA reductions in regions prior to visible tumour recur‐
rence [35].
Therefore, DTI has been proposed to improve the identification of peritumour infil‐
tration by better characterising diffusion abnormalities. Several studies have used the p
(isotropic) and q (anisotropic) maps which are derived from the diffusion tensor to iden‐
tify tumour and peritumour infiltration as it is proposed that the tumour core has a re‐
duced q and increased p and the infiltrative tumour has an increased p and unchanged q.
Defining the infiltrative tumour with p and q maps has been suggested as a replacement
for defining the CTV on standard anatomic imaging in order to reduce the overall radia‐
tion dose in radiotherapy plans [24]. The CTV will typically need to be manually drawn;
however, a study has proposed a semi‐automated method that automatically expands the
GTV drawn on the q map using a level set function on a p map [15].
Another proposed method utilises FA maps to detect tumour infiltration, as FA has
been found to be reduced in infiltrative regions. Study groups have used deep learning to
correct FA maps for free water to increase FA maps’ accuracy, finding that a model utilis‐
ing these corrected FA maps could predict recurrence with an area under the curve (AUC)
of 0.9 (compared to 0.77 with uncorrected maps) in a retrospective study [27]. The group
additionally used the method to create the CTV and found that it correlated with the typ‐
ical method of CTV definition, although it was larger in volume and adequately covered
reoccurrences in follow‐up scans [28].
A limitation of using p and q maps or FA maps to expand the GTV is that they do
not take directionality into account, and information on fibre orientation is not used. Some
studies have used the whole tensor to predict tumour spread. An article used a 3D‐based
Appl. Sci. 2022, 12, 816 10 of 16

DTI model that uses the fibre orientation to predict glioma growth [25]. However, this
study did not evaluate the model accuracy quantitatively. Another article used white mat‐
ter path length function to map the shortest path along the WM from a given region back
to the tumour core and expanded the GTV from this by 1, 2, or 3 cm to form an anisotropic
CTV. However, this method is only displayed in two cases, and the results of a retrospec‐
tive study using the software are awaited [26]. The group additionally released the soft‐
ware to be used openly, therefore making translational studies using the method much
easier.
Figure 5 shows a diagram showing how a DTI‐guided anisotropic CTV could theo‐
retically capture more areas of tumoral spread in gliomas by expanding along the WM.

Figure 5. Diagram showing glioma cells migrating along white matter tracts and how a DTI‐based
anisotropic clinical target volume (CTV) could better capture areas of tumoral spread than isotropic
CTV and potentially reduce the overall radiation dose.

3.3. Applications in Gliomas

The most promising use of DTI in gliomas as discussed is to define the CTV in areas
of tumour infiltration not visible on anatomical MR sequences or to predict areas of tu‐
mour reoccurrence. Several studies already discussed have developed methods to per‐
form this with one getting an AUC accuracy of 0.9 for predicting tumour recurrence sites
[27]. These methods have been applied to producing SRT treatment CTVs. Virtual dissec‐
tion would additionally be of benefit for gliomas in order to reduce radiation dose to the
eloquent WM tracts with a small number of theoretical SRT planning studies making use
of this technique, showing that it would be effective [9,29]. Unfortunately, there have not
been any prospective studies using a DTI‐guided SRT treatment plan to treat patients with
gliomas, as found by a recent systematic review [19]. As far as the authors are aware, there
have not been studies on DTI‐guided radiotherapy in any primary brain malignancies
other than gliomas [19], although there are studies using DTI to differentiate gliomas from
other primary brain malignancies, which is an important step for management planning
[16,17].

3.4. Applications in Brain Metastases

Brain metastases develop from the haematogenous spread of cancer cells and unlike
gliomas are typically not infiltrative [36,37]. Therefore, the main use of DTI in radiother‐
apy treatment planning would be to avoid salient WM tracts by including them as OAR.
Appl. Sci. 2022, 12, 816 11 of 16

This method has been evaluated in several studies of cohorts including brain metastasis
along with other benign neurological conditions. The largest identified study contained
20 patients with brain metastasis (and three patients with other conditions), and they com‐
pared SRT plans with and without including the CST as an OAR, finding that this signif‐
icantly reduced the maximum radiation dose to the CST but not the total dose to 95% of
the tract [21]. However, this study did not use the tractography‐guided treatment plan for
administration of the dose to the patients. There are studies that include patients with
metastasis that use radiotherapy‐guided treatment plans for the patients’ therapy, with
all of the studies finding that the maximum dose was significantly reduced to the dis‐
sected tracts [20,23,30]. These studies found no complications in the patients during short
follow‐up periods (all < 1.5 years); however, the subject numbers were small (three to nine
subjects each), and no studies included a control group. However, these studies show the
feasibility on integrating tractography into SRT treatment of brain metastases.

3.5. Application in Benign Conditions

SRT can be used in a range of non‐malignant neurological conditions, and therefore,
reducing the dose to nearby eloquent WM tracts to the pathology would likely reduce the
risk of complications. DTI integration is most widely explored in the benign condition,
arteriovenous malformation (AVM). A control study included 24 AVM patients, which
used tractography in SRT treatment planning and 28 subjects with AVM without using
tractography [22]. The subjects were followed up for 4 years, and the study found there
were no significant differences in the success of treatment on angiography and that the
use of tractography to model the CST significantly reduced the risk of motor complica‐
tions. A limitation of this study is that it was retrospective with the control group being
subjects from before the institution introduced tractography into SRT treatment planning.
Therefore, the groups were not randomised, and the improved outcomes may be due to
other factors such as improved operator skill over time. The same institution released a
review of all patients they treated for AVM since the routine implementation of tractog‐
raphy with SRT (from 2004 to 2009), which included 144 patients. They found that 46% of
cases utilised tractography, showing that it can be introduced effectively to a clinical set‐
ting, and a large proportion of patients may benefit from its integration. However, the use
of tractography integration was operator dependent and may have not been necessary in
some of the patients it was used in—or necessary in some of the cases when not integrated.
Although it was found that in 60% of cases utilising tractography, the tract was within 5
mm [31].
DTI‐integrated SRT has been explored in a small number of other neurological con‐
ditions, which include meningiomas and vestibular schwannomas. These studies include
cohorts with multiple conditions and find that integrating tractography reduces the max‐
imum radiation dose to the eloquent tracts [20,23]. However, the evidence for tractog‐
raphy implementation in the individual conditions is limited.

3.6. Future Considerations

The most important future work needed is randomised control trials on DTI‐guided
IMRT and SRT, as currently, none have been identified, although there is a non‐random‐
ised trial on tractography use in AVM [22]. The PRaM‐GBM study, which is aiming to
recruit 120 patients in the UK and performing DTI prior to surgery and IMRT of glioblas‐
tomas, is currently ongoing and promising [38]. However, the outline of this study is for
it to be observational, and therefore, it likely will not use DTI‐guided therapy plans in
treatment. At the same time, the study should give important information on using DTI
to predict areas of recurrence.
Work is additionally needed on the combined use of DTI‐guided CTV definition in
gliomas and inclusion of dissected tracts such as OAR, and software that optimises both
simultaneously may be required. Additionally, further work is needed on the extent to
which the CTV should be expanded along WM tracts, as it could potentially be expanded
Appl. Sci. 2022, 12, 816 12 of 16

further than the typical 2 cm and potentially reduce recurrence rates; Jordan et al. 2019
have developed software that is able to perform this. The study produced CTVs that were
expanded by 1, 2, and 3 cm along WM tracts; however, it does not compare the potential
effectiveness of each, as this would likely require a controlled trial [26].
Subjectivity may be able to be reduced in the future by using automated whole brain
tractography, which automatically dissects multiple eloquent tracts. This has previously
been shown to be useful in tractography‐guided stereotactic neurosurgery [39]. It may
additionally increase accuracy by including tracts not immediately thought to be near to
the pathology due to distorted anatomy. It may also reduce the doses to contralateral WM
tracts in large lesions such as those shown in Altabella et al. 2018 on DTI‐guided SRT plans
[9]. One limitation of automated whole brain tractography is that there is still subjectivity
regarding which tracts are included, and atlas‐based approaches may be prone to errors,
particularly in the presence of anatomy‐distorting lesions. However, this method may
play a role in reducing the time required for pre‐radiotherapy planning and increasing
standardisation, although the tracts would still need to be quality controlled by a highly
skilled practitioner.
There is additionally much progress in automated lesion segmentation due to im‐
provements in deep learning. Automated lesion segmentation may be able to be utilised
to delineate the initial GTV and previously discussed algorithms used to expand it to the
CTV. This could additionally remove subjectivity and speed up the SRT planning process.
An idealised fully automated pipeline for DTI‐guided radiotherapy planning is shown
in Figure 6, which could potentially be achieved with further research as suggested.

Figure 6. A proposed idealised pipeline for DTI‐guided SRT planning, which would be fully auto‐
mated.
Appl. Sci. 2022, 12, 816 13 of 16

4. Limitations
The biggest limitation for using DTI to guide radiotherapy planning is the lack of
controlled trials. The only controlled trial the authors, and a recent systematic review,
have identified is in AVM, and this is a non‐randomised retrospective study [22]. This
poses an issue, as while the majority of the SRT pre‐treatment plans using DTI guidance
deliver the same dose to the pathology, this does not necessarily mean the treatment will
have the same success rate on follow up. Additionally, while reduced radiation doses
overall to eloquent WM tracts will likely mean reduced complications, control trials are
still highly desirable.
An issue posed by gliomas is that due to infiltration along WM tracts, if it infiltrates
into an eloquent tract that is included as an OAR, then the dose may be reduced and in‐
crease the risk of recurrence. Therefore, further research in gliomas is required to evaluate
when and when not to include WM tracts as OAR.
An additional issue with gliomas is that their infiltration leads to reductions in FA,
which may reduce the tractography algorithm’s accuracy. This is due to tractography al‐
gorithms typically using an FA threshold, with tracts not including voxels below the
threshold. This could be accounted for by reducing the FA threshold, but this may lead to
an increase in false positive tracts. These downsides can be overcome using a combination
of more advanced diffusion acquisitions (i.e., with more directions and b‐values), diffu‐
sion models (e.g., spherical deconvolution), and probabilistic tractography.
Another limitation is that the accuracy of the dissected tracts is dependent on the
registration of the MRI images to the planning CT, as CT is used for the radiotherapy
planning. CT MRI fusion is more difficult than simply registering different MRI se‐
quences, and quality control is likely difficult due to different tissue contrasts. A recent
systematic review on the topic found many different methods for CT/MRI registration
with no standardisation and poor comparability of the validation of methods [40]. Alt‐
hough this article was assessing the issue in Stereotactic Electroencephalography, and
there are only a small number of radiotherapy treatment planning systems that have their
own in‐built co‐registration algorithms.
A practical limitation is that the integration of dissected WM tracts into the SRT treat‐
ment workstation may be difficult. Gavin et al. 2016 give a detailed step‐by‐step example
of how they integrated the tractography results into their ‘GammaPlan’ workstation [20];
however, due to varying software, standardisation will likely be difficult.
A consideration also required is the additional resources and time needed. Logistics
may be particularly difficult, as the DTI MRI scan will need to happen shortly before the
SRT to reduce error from brain changes over time and that DTI requires incredibly com‐
putationally complex processing, which can take a long time. DTI application will addi‐
tionally require highly skilled both technical and clinical operators to apply it correctly.
The decision of which tracts to include is additionally highly subjective, as this is operator‐
dependent. However, despite these limitations, an institution in Japan introduced tractog‐
raphy prior to the SRT of AVMs into routine clinical practice, using it in almost 50% of
cases. One way the group achieved this is by performing the DTI sequences the day prior
[22].

5. Conclusions
DTI‐guided radiotherapy is particularly promising for the modelling of eloquent
tracts using tractography to reduce the radiation dose by their inclusion as OAR in radio‐
therapy treatment planning. This has already been shown to improve motor complica‐
tions in a case‐controlled trial and is reported to be routinely clinically implemented as
part of SRT treatment of AVMs in one institution, and therefore, it could potentially be
referred to as a state‐of‐the‐art method. However, replication and further case‐controlled
clinical trials using this method are of a high priority before this method is used routinely,
especially in conditions other than AVM. There is additionally considerable promise in
Appl. Sci. 2022, 12, 816 14 of 16

using DTI to produce anisotropic CTV in treatment of gliomas. This could potentially im‐
prove current radiotherapy practice, as it may not only reduce the radiation dose but has
the potential to more accurately target invasive tumours. However, research into this is in
the initial phase, as there are currently only theoretical planning studies published with
no consensus on the method of optimally delineating the CTV boundary. Therefore, fur‐
ther theoretical studies and case control trials are required before it can be considered
viable for inclusion into clinical practice.

Author Contributions: Conceptualisation, J.C. and S.B.; writing—original draft preparation, J.C.;
writing—review and editing, L.M., S.M., M.G., M.K. and S.B.; creation of figures, J.C.; supervision,
S.B. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Halasz, L.M.; Rockhill, J.K. Stereotactic radiosurgery and stereotactic radiotherapy for brain metastases. Surg. Neurol. Int. 2013,
4, S185. https://doi.org/10.4103/2152‐7806.111295.
2. Niyazi, M.; Brada, M.; Chalmers, A.J.; Combs, S.E.; Erridge, S.C.; Fiorentino, A.; Grosu, A.L.; Lagerwaard, F.J.; Minniti, G.;
Mirimanoff, R.O.; et al. ESTRO‐ACROP guideline “target delineation of glioblastomas.” Radiother. Oncol. 2016, 118, 35–42.
https://doi.org/10.1016/J.RADONC.2015.12.003.
3. Taylor, A.; Powell, M.E.B. Intensity‐modulated radiotherapy—what is it? Cancer Imaging 2004, 4, 68–73.
https://doi.org/10.1102/1470‐7330.2004.0003.
4. Mamalui‐Hunter, M.; Yaddanapudi, S.; Zhao, T.; Mutic, S.; Low, D.A.; Drzymala, R.E. Patient‐specific independent 3D Gam‐
maPlan quality assurance for Gamma Knife Perfexion radiosurgery. J. Appl. Clin. Med. Phys. 2013, 14, 62–70.
https://doi.org/10.1120/JACMP.V14I1.3949.
5. Burnet, N.G.; Thomas, S.J.; Burton, K.E.; Jefferies, S.J. Defining the tumour and target volumes for radiotherapy. Cancer Imaging
2004, 4, 153–161. https://doi.org/10.1102/1470‐7330.2004.0054.
6. Armstrong, C.L.; Gyato, K.; Awadalla, A.W.; Lustig, R.; Tochner, Z.A. A Critical Review of the Clinical Effects of Therapeutic
Irradiation Damage to the Brain: The Roots of Controversy. Neuropsychol. Rev. 2004, 14, 65–86.
https://doi.org/10.1023/B:NERV.0000026649.68781.8E.
7. Dhermain, F.; Barani, I.J. Complications from radiotherapy. Handb. Clin. Neurol. 2016, 134, 219–234. https://doi.org/10.1016/B978‐
0‐12‐802997‐8.00013‐X.
8. Chapman, C.H.; Zhu, T.; Nazem‐Zadeh, M.; Tao, Y.; Buchtel, H.A.; Tsien, C.I.; Lawrence, T.S.; Cao, Y. Diffusion tensor imaging
predicts cognitive function change following partial brain radiotherapy for low‐grade and benign tumors. Radiother. Oncol.
2016, 120, 234–240. https://doi.org/10.1016/J.RADONC.2016.06.021.
9. Altabella, L.; Broggi, S.; Mangili, P.; Conte, G.M.; Pieri, V.; Iadanza, A.; del Vecchio, A.; Anzalone, N.; di Muzio, N.; Calandrino,
R.; et al. Integration of Diffusion Magnetic Resonance Tractography into tomotherapy radiation treatment planning for high‐
grade gliomas. Phys. Med. 2018, 55, 127–134. https://doi.org/10.1016/J.EJMP.2018.10.004.
10. Price, S.J.; Gillard, J.H. Imaging biomarkers of brain tumour margin and tumour invasion. Br. J. Radiol. 2011, 84, S159.
https://doi.org/10.1259/BJR/26838774.
11. Trip, A.K.; Jensen, M.B.; Kallehauge, J.F.; Lukacova, S. Individualizing the radiotherapy target volume for glioblastoma using
DTI‐MRI: A phase 0 study on coverage of recurrences. Acta Oncol. 2019, 58, 1532–1535.
https://doi.org/10.1080/0284186X.2019.1637018.
12. O’Donnell, L.J.; Westin, C.‐F. An introduction to diffusion tensor image analysis. Neurosurg. Clin. N. Am. 2011, 22, 185.
https://doi.org/10.1016/J.NEC.2010.12.004.
13. Peña, A.; Green, H.A.L.; Carpenter, T.A.; Price, S.J.; Pickard, J.D.; Gillard, J.H. Enhanced visualization and quantification of
magnetic resonance diffusion tensor imaging using the p:q tensor decomposition. Br. J. Radiol. 2014, 79, 101–109.
https://doi.org/10.1259/BJR/24908512.
14. Razek, A.A.K.A.; El‐Serougy, L.; Abdelsalam, M.; Gaballa, G.; Talaat, M. Differentiation of residual/recurrent gliomas from
postradiation necrosis with arterial spin labeling and diffusion tensor magnetic resonance imaging‐derived metrics. Neuroradi‐
ology 2018, 60, 169–177. https://doi.org/10.1007/S00234‐017‐1955‐3.
15. Rahmat, R.; Brochu, F.; Li, C.; Sinha, R.; Price, S.J.; Jena, R. Semi‐automated construction of patient individualised clinical target
volumes for radiotherapy treatment of glioblastoma utilising diffusion tensor decomposition maps. Br. J. Radiol. 2020, 93, 93.
https://doi.org/10.1259/BJR.20190441.
Appl. Sci. 2022, 12, 816 15 of 16

16. Thong, P.M.; Duc, N.M. The Role of Apparent Diffusion Coefficient in the Differentiation between Cerebellar Medulloblastoma
and Brainstem Glioma. Neurol. Int. 2020, 12, 34–40. https://doi.org/10.3390/NEUROLINT12030009.
17. Duc, N.M. The role of diffusion tensor imaging metrics in the discrimination between cerebellar medulloblastoma and brain‐
stem glioma. Pediatr. Blood Cancer 2020, 67, e28468. https://doi.org/10.1002/PBC.28468.
18. Essayed, W.I.; Zhang, F.; Unadkat, P.; Cosgrove, G.R.; Golby, A.J.; O’Donnell, L.J. White matter tractography for neurosurgical
planning: A topography‐based review of the current state of the art. NeuroImage Clin. 2017, 15, 659–672.
https://doi.org/10.1016/J.NICL.2017.06.011.
19. Yahya, N.; Manan, H.A. Utilisation of Diffusion Tensor Imaging in Intracranial Radiotherapy and Radiosurgery Planning for
White Matter Dose Optimization: A Systematic Review. World Neurosurg. 2019, 130, e188–e198.
https://doi.org/10.1016/J.WNEU.2019.06.027.
20. Gavin, C.G.; Sabin, H.I. Stereotactic diffusion tensor imaging tractography for Gamma Knife radiosurgery. J. Neurosurg. 2016,
125, 139–146. https://doi.org/10.3171/2016.8.GKS161032.
21. Kawasaki, K.; Matsumoto, M.; Kase, M.; Nagano, O.; Aoyagi, K.; Kageyama, T. Quantification of the radiation dose to the py‐
ramidal tract using tractography in treatment planning for stereotactic radiosurgery. Radiol. Phys. Technol. 2017, 10, 507–514.
https://doi.org/10.1007/S12194‐017‐0411‐8.
22. Koga, T.; Shin, M.; Maruyama, K.; Kamada, K.; Ota, T.; Itoh, D.; Kunii, N.; Ino, K.; Aoki, S.; Masutani, Y.; et al. Integration of
Corticospinal Tractography Reduces Motor Complications After Radiosurgery. Int. J. Radiat. Oncol. 2012, 83, 129–133.
https://doi.org/10.1016/J.IJROBP.2011.05.036.
23. Sun, L.; Qu, B.; Wang, J.; Ju, Z.; Zhang, Z.; Cui, Z.; Jack, Y.; Ling, Z.; Yu, X.; Pan, L. Integration of Functional MRI and White
Matter Tractography in CyberKnife Radiosurgery. Technol. Cancer Res. Treat. 2017, 16, 850–856.
https://doi.org/10.1177/1533034617705283.
24. Berberat, J.; McNamara, J.; Remonda, L.; Bodis, S.; Rogers, S. Diffusion tensor imaging for target volume definition in glioblas‐
toma multiforme. Strahlenther. Onkol. 2014, 190, 939–943. https://doi.org/10.1007/S00066‐014‐0676‐3.
25. Hathout, L.; Patel,V.; Wen, P. A 3‐dimensional DTI MRI‐based model of GBM growth and response to radiation therapy. Int. J.
Oncol. 2016, 49, 1081–1087. https://doi.org/10.3892/IJO.2016.3595.
26. Jordan, K.; Morin, O.; Wahl, M.; Amirbekian, B.; Chapman, C.; Owen, J.; Mukherjee, P.; Braunstein, S.; Henry, R. An Open‐
Source Tool for Anisotropic Radiation Therapy Planning in Neuro‐oncology Using DW‐MRI Tractography. Front. Oncol. 2019,
9, 810. https://doi.org/10.3389/FONC.2019.00810.
27. Metz, M.‐C.; Molina‐Romero, M.; Lipkova, J.; Gempt, J.; Liesche‐Starnecker, F.; Eichinger, P.; Grundl, L.; Menze, B.; Combs, S.E.;
Zimmer, C.; et al. Predicting Glioblastoma Recurrence from Preoperative MR Scans Using Fractional‐Anisotropy Maps with
Free‐Water Suppression. Cancers 2020, 12, 728. https://doi.org/10.3390/CANCERS12030728.
28. Peeken, J.C.; Molina‐Romero, M.; Diehl, C.; Menze, B.H.; Straube, C.; Meyer, B.; Zimmer, C.; Wiestler, B.; Combs, S.E. Deep
learning derived tumor infiltration maps for personalized target definition in Glioblastoma radiotherapy. Radiother. Oncol. 2019,
138, 166–172. https://doi.org/10.1016/J.RADONC.2019.06.031.
29. Igaki, H.; Sakumi, A.; Mukasa, A.; Saito, K.; Kunimatsu, A.; Masutani, Y.; Hanakita, S.; Ino, K.; Haga, A.; Nakagawa, K.; et al.
Corticospinal tract‐sparing intensity‐modulated radiotherapy treatment planning. Rep. Pr. Oncol. Radiother. 2014, 19, 310–316.
https://doi.org/10.1016/J.RPOR.2014.01.002.
30. Conti, A.; Pontoriero, A.; Ricciardi, G.K.; Granata, F.; Vinci, S.; Angileri, F.F.; Pergolizzi, S.; Alafaci, C.; Rizzo, V.; Quartarone,
A.; et al. Integration of functional neuroimaging in CyberKnife radiosurgery: Feasibility and dosimetric results. Neurosurg. Focus
2013, 34, E5. https://doi.org/10.3171/2013.2.FOCUS12414.
31. Koga, T.; Maruyama, K.; Kamada, K.; Ota, T.; Shin, M.; Itoh, D.; Kunii, N.; Ino, K.; Terahara, A.; Aoki, S.; et al. Outcomes of
Diffusion Tensor Tractography–Integrated Stereotactic Radiosurgery. Int. J. Radiat. Oncol. 2012, 82, 799–802.
https://doi.org/10.1016/J.IJROBP.2010.11.046.
32. Mair, D.B.; Ames, H.M.; Li, R. Mechanisms of invasion and motility of high‐grade gliomas in the brain. Mol. Biol. Cell 2018, 29,
2509–2515. https://doi.org/10.1091/MBC.E18‐02‐0123.
33. Engwer, C.; Hillen, T.; Knappitsch, M.; Surulescu, C. Glioma follow white matter tracts: A multiscale DTI‐based model. J. Math.
Biol. 2014, 71, 551–582. https://doi.org/10.1007/S00285‐014‐0822‐7.
34. Lu, S.; Ahn, D.; Johnson, G.; Law, M.; Zagzag, D.; Grossman, R.I. Diffusion‐Tensor MR Imaging of Intracranial Neoplasia and
Associated Peritumoral Edema: Introduction of the Tumor Infiltration Index. Radiology 2004, 232, 221–228.
https://doi.org/10.1148/RADIOL.2321030653.
35. Jin, Y.; Randall, J.W.; Elhalawani, H.; Feghali, K.A. Al; Elliott, A.M.; Anderson, B.M.; Lacerda, L.; Tran, B.L.; Mohamed, A.S.;
Brock, K.K.; et al. Detection of Glioblastoma Subclinical Recurrence Using Serial Diffusion Tensor Imaging. Cancers 2020, 12,
568. https://doi.org/10.3390/CANCERS12030568.
36. Pavlisa, G.; Rados, M.; Pavlisa, G.; Pavic, L.; Potocki, K.; Mayer, D. The differences of water diffusion between brain tissue
infiltrated by tumor and peritumoral vasogenic edema. Clin. Imaging 2009, 33, 96–101. https://doi.org/10.1016/J.CLIN‐
IMAG.2008.06.035.
37. Eichler, A.F.; Chung, E.; Kodack, D.P.; Loeffler, J.S.; Fukumura, D.; Jain, R.K. The biology of brain metastases—translation to
new therapies. Nat. Rev. Clin. Oncol. 2011, 8, 344. https://doi.org/10.1038/NRCLINONC.2011.58.
38. Predicting Sites of Tumour Progression in the Invasive Margin of Glioblastomas (PRaM‐GBM Study)—Full Text View—Clini‐
calTrials.gov. Available online: https://clinicaltrials.gov/ct2/show/NCT03294434 (accessed on 6 October 2021).
 Appl. Sci. 2022, 12, 816 16 of 16

39. Zakaria, H.; Haider, S.; Lee, I. Automated Whole Brain Tractography Affects Preoperative Surgical Decision Making. Cureus
2017, 9, e1656. https://doi.org/10.7759/CUREUS.1656.
40. Perez, J.; Mazo, C.; Trujillo, M.; Herrera, A. MRI and CT Fusion in Stereotactic Electroencephalography: A Literature Review.
Appl. Sci. 2021, 11, 5524. https://doi.org/10.3390/APP11125524.

View publication stats