Papers by Alessandro Riccardi
The CAD system comprises at present four macro-steps [Riccardi06]: 1. Lung segmentation; 2. 2D no... more The CAD system comprises at present four macro-steps [Riccardi06]: 1. Lung segmentation; 2. 2D nodule-like signals detection in slices: logical AND of a. Fast Radial filter; b. Scale Space filter; 3. Grouping of 2D signals into 3D candidate nodules; 4. False Positive Reduction (FPR): a. Coarse FPR (on the basis of a few geometrical features); b. Fine FPR: i. 2D Gray Level features and Support Vector Machine classifier; ii. 3D Ranklet-Based features and Support Vector Machine classifier; iii. 2D Support Vector Regression Filtering FPR. Steps i, ii, iii can be combined in different modalities.
International Journal of Modern Physics C, 2002
EGS is a very popular Monte Carlo code, used in the simulation of Nuclear Medicine devices. Simul... more EGS is a very popular Monte Carlo code, used in the simulation of Nuclear Medicine devices. Simulation techniques are particularly effective to optimize collimator configuration and camera design in Single Photon Emission studies. With the EGS code, users must define the geometry where particles are transported. This can be both a very hard task and a source of inefficiency, especially in the case of complex geometries as, for instance, hexagonal hole collimators or pixellated detectors. In this paper we present a modular description of such geometries. Our method allows the computation of the region a point belongs to in a few steps; thus we are able to calculate this region in a reduced number of operations, independently of the collimator and detector dimensions. With a modular description we can reduce the computational time by 30%, with respect to a "traditional" description of the geometry. We validated the modular description in the simulation of a Nuclear Medicine ...
Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 2003
Scintimammography shows strong potential in detecting and differentiating breast cancer. This sci... more Scintimammography shows strong potential in detecting and differentiating breast cancer. This scintigraphic technique, using a standard gamma camera, allows high sensitivity and specificity values (>95%) for detected tumors more than 1 cm size. However, the sensitivity of scintimammography using conventional gamma cameras is considerably less (40-50%) for tumors with smaller size. Recently, the authors demonstrated how the use of a small FOV dedicated gamma camera (Single Photon Emission Mammography, or SPEM camera), with very high intrinsic spatial resolution (1.7 mm FWHM), working with breast moderately compressed and positioned close to the breast tumor (i.e., analogously to X-ray mammography) increased sensitivity up to 80% for tumors sized between 0.5 and 1 cm (T1b). The aim of this paper is to demonstrate how the reduced breast thickness can play a primary role in small cancer detection. Five different methods were taken into account: clinical measurements, comparing tumor SNR values obtained from the same patients in prone scintimammography and in SPEM, comparing SNR values between compressed and uncompressed breast in craniocaudal projection, breast phantom measurements, Monte Carlo simulations and simplified theoretical model. Results confirm that the mechanism for the improvement in visualizing sub-centimeter lesions due to compression is a reduction of lesion-detector distance. As a result of this reduced distance there is a less reabsorption of signal by interposed breast tissue, and improved detector intrinsic spatial resolution.
Journal of Electronic Imaging, 1999
The existence of clustered microcalcifications is one of the important early signs of breast canc... more The existence of clustered microcalcifications is one of the important early signs of breast cancer. This paper presents an image processing procedure for the automatic detection of clustered microcalcifications in digitized mammograms. In particular, a sensi- tivity range of around one false positive per image is targeted. The proposed method consists of two main steps. First, possible micro- calcification pixels in the mammograms are segmented out using wavelet features or both wavelet features and gray level statistical features, and labeled into potential individual microcalcification ob- jects by their spatial connectivity. Second, individual microcalcifica- tions are detected by using the structure features extracted from the potential microcalcification objects. The classifiers used in these two steps are feedforward neutral networks. The method is applied to a database of 40 mammograms (Nijmegen database) containing 105 clusters of microcalcifications. A free response operating character- istics curve is used to evaluate the performance. Results show that the proposed procedure gives quite satisfactory detection perfor- mance. In particular, a 93% mean true positive detection rate is achieved at the price of one false positive per image when both wavelet features and gray level statistical features are used in the first step. © 1999 SPIE and IS&T. (S1017-9909(99)00701-1)
Digital Mammography, 2003
In this paper we present a novel approach to mass detection in digital mammograms. The great vari... more In this paper we present a novel approach to mass detection in digital mammograms. The great variability of the masses appearance is the main obstacle of building a mass detection method. It is indeed demanding to characterize all the varieties of masses with a reduced set of features. Hence, in our approach we decide not to extract any feature, for the detection of the region of interest; on the contrary we exploit all the information available on the image. No a priori knowledge and no appearance model are used. A multiresolution overcomplete wavelet representation is achieved, in order to codify the image with redundancy of information. The vectors of the very-large space obtained are classified by means of an SVM classifier. Training, validation and test are accomplished on images coming from USF DDSM database. The sensitivity of the presented system is 84% with a false-positive rate of 3.1 marks per image.
Reduction of False Positive signals (FPR) is a fundamental, yet awkward, step in computer aided m... more Reduction of False Positive signals (FPR) is a fundamental, yet awkward, step in computer aided mass detection schemes. This paper describes preliminary results of a filtering approach to FPR based on Support Vector Regression (SVR), a machine learning algorithm arising from a well-founded theoretical framework, the Statistical Learning Theory, which has recently proved to be superior to the conventional Neural Network framework for both classification and regression tasks: indeed, the proposed filtering method belongs to the family of neural filters. The SVR filter is forced to associate subregions extracted from input images, masses and non-masses, to continuous output values ranging from 0 to 1 representing a measure of the presence in the subregion of a mass. A weighted sum of outputs over each image is used to accomplish the FPR task. In the test phase, this approach reached promising results, retaining 87% of masses while reducing False Positives to 62%.
Physics in Medicine and Biology, 2001
In this work, we present a novel approach to mass detection in digital mammograms. The great vari... more In this work, we present a novel approach to mass detection in digital mammograms. The great variability of the masses appearance is the main obstacle of building a mass detection method. It is indeed demanding to characterize all the varieties of masses with a reduced set of features. Hence, in our approach we have chosen not to extract any feature, for the detection of the region of interest; on the contrary, we exploit all the information available on the image. A multiresolution overcomplete wavelet representation is performed, in order to codify the image with redundancy of information. The vectors of the very-large space obtained are then provided to a first SVM classifier. The detection task is here considered as a two-class pattern recognition problem: crops are classified as suspect or not, by using this SVM classifier. False candidates are eliminated with a second cascaded SVM. To further reduce the number of false positives, an ensemble of experts is applied: the final su...
This work describes the development of a new Computer Aided system (CAD) for the detection of nod... more This work describes the development of a new Computer Aided system (CAD) for the detection of nodules in CT scans of the lung, employing Computer Vision and Pattern Recognition techniques. The system consists of three steps: pre-processing, filtering and False Positive Reduction (FPR). The pre-processing step is dedicated to lung segmentation, and it is mainly based on Gray Level Histogram Thresholding, Seeded Region Growing and Mathematical Morphology. The second and third steps are aimed at detecting nodule-like signals – the filtering step-and at separating these signals into true and false nodules-the FPR step. The main characteristics of the CAD system are: 1) an original and iterative use of the Fast Radial filter, able to detect signals with circular symmetry in CT images; 2) the novel use of a filter based on the Scale-Space theory, able to locate circular signals of a given size, 3) the logical AND of the previous two filters. The iterative application of the Fast Radial fi...
2002 IEEE Nuclear Science Symposium Conference Record, 2003
A prototype for positron emission mammography is under development within a collaboration of the ... more A prototype for positron emission mammography is under development within a collaboration of the Italian Universities of Pisa, Ferrara, Bologna and Roma. The device is composed of two stationary detection heads, each with an active area of 6 cm × 6 cm, made of 30×30 YAP:Ce finger crystals of 2 mm × 2 mm × 30 mm. The EGSnrc Monte
Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 2004
A prototype for Positron Emission Mammography, the YAP-PEM, is under development within a collabo... more A prototype for Positron Emission Mammography, the YAP-PEM, is under development within a collaboration of the Italian Universities of Pisa, Ferrara, and Bologna. The aim is to detect breast lesions, with dimensions of 5 mm in diameter, and with a specific activity ratio of 10:1 between the cancer and breast tissue. The YAP-PEM is composed of two stationary detection heads of 6 Â 6 cm 2 , composed of a matrix of 30 Â 30 YAP:Ce finger crystals of 2 Â 2 Â 30 mm 3 each. The EGSnrc Monte Carlo code has been used to simulate several characteristics of the prototype. A fast EM algorithm has been adapted to reconstruct all of the collected lines of flight, also at large incidence angles, by achieving 3D positioning capability of the lesion in the FOV. The role of the breast compression has been studied. The performed study shows that a 5 mm diameter tumor of 37 kBq/cm 3 (1 mCi/cm 3), embedded in active breast tissue with 10:1 tumor/ background specific activity ratio, is detected in 10 min with a Signal-to-Noise Ratio of 8.771.0. Two hot lesions in the active breast phantom are clearly visible in the reconstructed image.
Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 2003
Compact gamma cameras based on arrays of compact Position Sensitive Photomultipliers (PSPMTs) (Ha... more Compact gamma cameras based on arrays of compact Position Sensitive Photomultipliers (PSPMTs) (Hamamatsu R7600-C8/12) were recently developed by several research groups. The previous generation of dedicated gamma cameras (5 in. PSPMT) demonstrated the clinical benefit and general diagnostic value for functional breast imaging in comparison with conventional nuclear medicine technique (Anger Camera prone scintimammography and 99m Tc Sestamibi administration). The aim of this paper is to investigate how scintillation material and pixel size of crystal arrays can improve image contrast and tumor SNR values. In this paper we compare tumor Signal-to-Noise Ratio (SNR) results obtained by imagers based on CsI(Tl) and NaI(Tl) array, respectively, by means of a liquid and solid breast phantom. The data collected by NaI(Tl) array show a improvement of SNR values for small tumor size (less than 8 mm). The improvement is also evident in small camera, even though for tumor size less than 6 mm the results are near visibility limit.
Medical Physics, 2011
The authors presented a novel system for automated nodule detection in lung CT exams. The approac... more The authors presented a novel system for automated nodule detection in lung CT exams. The approach is based on (1) a lung tissue segmentation preprocessing step, composed of histogram thresholding, seeded region growing, and mathematical morphology; (2) a filtering step, whose aim is the preliminary detection of candidate nodules (via 3D fast radial filtering) and estimation of their geometrical features (via scale space analysis); and (3) a false positive reduction (FPR) step, comprising a heuristic FPR, which applies thresholds based on geometrical features, and a supervised FPR, which is based on support vector machines classification, which in turn, is enhanced by a feature extraction algorithm based on maximum intensity projection processing and Zernike moments. The system was validated on 154 chest axial CT exams provided by the lung image database consortium public database. The authors obtained correct detection of 71% of nodules marked by all radiologists, with a false positive rate of 6.5 false positives per patient (FP/patient). A higher specificity of 2.5 FP/patient was reached with a sensitivity of 60%. An independent test on the ANODE09 competition database obtained an overall score of 0.310. The system shows a novel approach to the problem of lung nodule detection in CT scans: It relies on filtering techniques, image transforms, and descriptors rather than region growing and nodule segmentation, and the results are comparable to those of other recent systems in literature and show little dependency on the different types of nodules, which is a good sign of robustness.
IEEE Transactions on Nuclear Science, 2003
The introduction of a new gamma camera fully dedicated to scintimammography (Single Photon Emissi... more The introduction of a new gamma camera fully dedicated to scintimammography (Single Photon Emission Mammography-SPEM), and more recently with a full breast FoV, allowed to make clinical examination in cranio-caudal projection like in RXmammography, with breast mildly compressed. Such cameras are based on pixellated scintillation array and position sensitive photomultiplier (PSPMT). Reducing the collimator-tumor distance, the geometric spatial resolution and contrast was enhanced. Unfortunately, due to the scintimammographic low counting, poor contrast images are still obtained, in particular for small tumor. The aim of this paper is to evaluate how a camera based on pixellated detector can improve the SNR values for small tumor by an effective correction of the spatial response. The procedure is based on good pixel identification. A Small Gamma Camera (SGC) was arranged using metal channel dynode PSPMT photomultiplier (Hamamatsu R7600-C8) coupled to different CsI (Tl) scintillator array, with field of view (FoV) with an all purpose collimator. This PSPMT kind drastically reduces the charge spread improving the intrinsic characteristics of the imager. The dimensions of the CsI (Tl) arrays were the same of PSPMT active area (22x22mm 2). Considering the very high intrinsic spatial resolution, a look up table was realized to accurately correct the gain and spatial non-uniformities. We used a breast and torso phantom to characterize the SNR as a function of scintillation pixel size, thickness of the breast, tumor size and depth. The data showed that the SNR depends principally on the match between the tumor and pixel size. In particular, for a 6 mm diameter tumor, the best SNR results were obtained by a 2x2 mm 2 pixelled array. For larger tumors, up to 10 mm diameter, a greater pixel size, like 3x3 mm 2 or 4x4 mm 2 , optimizes the SNR value. We compared the results of this camera with the analogous ones obtained by a SPEM gamma camera and by a standard Anger Camera.
IEEE Transactions on Nuclear Science, 2004
The capability of the scintimammography to diagnose subcentimeters sized tumors was increased by ... more The capability of the scintimammography to diagnose subcentimeters sized tumors was increased by the employment of a dedicated gamma camera. The introduction of small field of view camera, based on pixellated scintillation array and position sensitive photomultiplier, allowed to enhance the geometric spatial resolution and contrast of the images due to reduced collimator-tumor distance. The aim of this paper is
This work describes the development of a new Computer Aided system (CAD) for the detection of nod... more This work describes the development of a new Computer Aided system (CAD) for the detection of nodules in CT scans of the lung, employing Computer Vision and Pattern Recognition techniques. The system consists of three steps: pre-processing, filtering and False Positive Reduction (FPR). The pre-processing step is dedicated to lung segmentation,
The authors presented a novel system for automated nodule detection in lung CT exams. The approac... more The authors presented a novel system for automated nodule detection in lung CT exams. The approach is based on (1) a lung tissue segmentation preprocessing step, composed of histogram thresholding, seeded region growing, and mathematical morphology; (2) a filtering step, whose aim is the preliminary detection of candidate nodules (via 3D fast radial filtering) and estimation of their geometrical features (via scale space analysis); and (3) a false positive reduction (FPR) step, comprising a heuristic FPR, which applies thresholds based on geometrical features, and a supervised FPR, which is based on support vector machines classification, which in turn, is enhanced by a feature extraction algorithm based on maximum intensity projection processing and Zernike moments. The system was validated on 154 chest axial CT exams provided by the lung image database consortium public database. The authors obtained correct detection of 71% of nodules marked by all radiologists, with a false positive rate of 6.5 false positives per patient (FP/patient). A higher specificity of 2.5 FP/patient was reached with a sensitivity of 60%. An independent test on the ANODE09 competition database obtained an overall score of 0.310. The system shows a novel approach to the problem of lung nodule detection in CT scans: It relies on filtering techniques, image transforms, and descriptors rather than region growing and nodule segmentation, and the results are comparable to those of other recent systems in literature and show little dependency on the different types of nodules, which is a good sign of robustness.
A prototype for Positron Emission Mammography, the YAP-PEM, is under development within a collabo... more A prototype for Positron Emission Mammography, the YAP-PEM, is under development within a collaboration of the Italian Universities of Pisa, Ferrara, and Bologna. The aim is to detect breast lesions, with dimensions of 5 mm in diameter, and with a specific activity ratio of 10:1 between the cancer and breast tissue. The YAP-PEM is composed of two stationary detection heads of 6 Â 6 cm 2 , composed of a matrix of 30 Â 30 YAP:Ce finger crystals of 2 Â 2 Â 30 mm 3 each. The EGSnrc Monte Carlo code has been used to simulate several characteristics of the prototype. A fast EM algorithm has been adapted to reconstruct all of the collected lines of flight, also at large incidence angles, by achieving 3D positioning capability of the lesion in the FOV. The role of the breast compression has been studied. The performed study shows that a 5 mm diameter tumor of 37 kBq/cm 3 (1 mCi/cm 3), embedded in active breast tissue with 10:1 tumor/ background specific activity ratio, is detected in 10 min with a Signal-to-Noise Ratio of 8.771.0. Two hot lesions in the active breast phantom are clearly visible in the reconstructed image.
The introduction of a new gamma camera fully dedicated to scintimammography (Single Photon Emissi... more The introduction of a new gamma camera fully dedicated to scintimammography (Single Photon Emission Mammography-SPEM), and more recently with a full breast FoV, allowed to make clinical examination in cranio-caudal projection like in RXmammography, with breast mildly compressed. Such cameras are based on pixellated scintillation array and position sensitive photomultiplier (PSPMT). Reducing the collimator-tumor distance, the geometric spatial resolution and contrast was enhanced. Unfortunately, due to the scintimammographic low counting, poor contrast images are still obtained, in particular for small tumor. The aim of this paper is to evaluate how a camera based on pixellated detector can improve the SNR values for small tumor by an effective correction of the spatial response. The procedure is based on good pixel identification. A Small Gamma Camera (SGC) was arranged using metal channel dynode PSPMT photomultiplier (Hamamatsu R7600-C8) coupled to different CsI (Tl) scintillator array, with field of view (FoV) with an all purpose collimator. This PSPMT kind drastically reduces the charge spread improving the intrinsic characteristics of the imager. The dimensions of the CsI (Tl) arrays were the same of PSPMT active area (22x22mm 2). Considering the very high intrinsic spatial resolution, a look up table was realized to accurately correct the gain and spatial non-uniformities. We used a breast and torso phantom to characterize the SNR as a function of scintillation pixel size, thickness of the breast, tumor size and depth. The data showed that the SNR depends principally on the match between the tumor and pixel size. In particular, for a 6 mm diameter tumor, the best SNR results were obtained by a 2x2 mm 2 pixelled array. For larger tumors, up to 10 mm diameter, a greater pixel size, like 3x3 mm 2 or 4x4 mm 2 , optimizes the SNR value. We compared the results of this camera with the analogous ones obtained by a SPEM gamma camera and by a standard Anger Camera.
Uploads
Papers by Alessandro Riccardi