arXiv subjects

Hossein Arabi

Publications and source records attributed to Hossein Arabi.

At least 19 recordsLinked to original sources

arXiv2025

Position-Sensitive Silicon Photomultiplier Array with Enhanced Position Reconstruction by means of a Deep Neural Network

Single-photon sensitive detectors like Silicon Photomultipliers are widely used in many medical imaging applications. By using detectors with position resolutions, it is possible to build compact photodetector readouts with reduced number of channels, but still preserving position resolution and gamma-rays imaging capabilities. In this work, we present the advantage of using a Deep Neural Networks (DNNs) light position reconstruction applied to a 2x2 array of linearly-graded SiPMs (LG-SiPMs), to minimize the distortions on the reconstructed event maps. Our approach significantly enhances both the resolution and linearity of position detection compared to the nominal reconstruction formula based on the device architecture. Remarkably, the DNN-based reconstruction boosts the number of resolved areas (pixels) by a factor of 5.7 to 12.1 (depending the training splitting used) allowing for a higher level of precision and performance in light detection.

physics.ins-det↗

arXiv2025

Evaluating the Impact of Partial Volume Correction on FDG PET Radiomics Reproducibility in Lymphoma Lesions

To evaluate how partial volume correction (PVC) affects the reproducibility of 18F-FDG PET radiomic features in lymphoma lesions, with respect to lesion volume and tissue type. This single-center retrospective study included 131 newly diagnosed lymphoma patients who underwent baseline 18F-FDG PET/CT. In total, 1,603 lesions (1,302 lymph nodes, 117 spleen/liver, 150 bone, and 34 bone/soft-tissue) were semi-automatically segmented and grouped by volume (<3, 3-10, 10-30, >30 mL) and tissue type. 93 radiomic features were extracted from non-PVC and PVC images processed with the Richardson-Lucy (RL) and Reblurred Van Cittert (RVC) algorithms following IBSI guidelines. Reproducibility was quantified using the coefficient of variation (CoV) and the intraclass correlation coefficient (ICC2, absolute agreement), with statistical comparisons performed via Mann-Whitney U tests and false-discovery-rate (FDR) correction. PVC significantly improved feature reproducibility, particularly for large lesions (>30 mL), with median ICC2>0.90 across most feature categories. Small lesions (<3 mL) showed lower stability (ICC2=0.84-0.94) and higher CoV (0.09-0.21). First-Order and GLCM features were the most robust overall (ICC2=0.92-0.99; CoV=0.07-0.11). Bone and spleen lesions exhibited the highest reproducibility (median ICC2 approx 0.95), whereas lymph-node and liver features were more variable. All volume- and tissue-dependent differences remained significant after FDR correction (p<0.05). PVC using RL and RVC markedly enhances FDG-PET radiomic reproducibility in lymphoma, particularly for larger and structurally uniform lesions. Robust features such as First-Order and GLCM can support standardized radiomics workflows and the development of reliable biomarkers for prognosis and personalized therapy. Multicenter validation is warranted to confirm generalizability beyond a single-center setting.

physics.med-ph↗

arXiv2025

Per-Lesion Radiomics Analysis of 68Ga-DOTA FAPI-46 and 18F-FDG PET/CT in Non-Small Cell Lung Cancer: A pilot Study

This pilot study compares per-lesion radiomics features of [68Ga]-DOTA FAPI-46 and [18F]-FDG PET/CT in non-small cell lung cancer (NSCLC) to explore complementary insights into intratumoral heterogeneity beyond conventional SUV metrics, aiming to enhance lesion characterization and clinical decision-making. A total of 28 PET/CT scans (14 [18F]-FDG and 14 [68Ga]-DOTA FAPI-46) were acquired for the initial staging of biopsy-confirmed NSCLC. A total of 81 co-localized lesions (lung: 21, mediastinal lymph nodes: 42, bone: 18) were segmented, with radiomics features extracted via PyRadiomics after IBSI-compliant preprocessing. Paired per-lesion comparisons used t-tests or Wilcoxon signed-rank tests with Benjamini-Hochberg FDR correction. Significant differences (adjusted P < 0.05) were identified across intensity, texture, and shape features. In lung lesions, FAPI showed lower first-order metrics but higher variance and GLCM contrast, suggesting stromal heterogeneity. Mediastinal lymph nodes had fewer differences, with FAPI exhibiting lower run percentage (GLRLM: -0.298, -6.196, P=1.45E-08). Bone lesions showed extensive variations, including reduced FAPI entropy (e.g., Entropy: -1.743, -5.798, P=6.95E-08). Feature overlaps highlighted complementary stromal (FAPI) and metabolic (FDG) insights. This pilot study demonstrates that per-lesion radiomics can capture complementary biological information from FAPI and FDG PET in NSCLC, highlighting intratumoral heterogeneity and stromal activity not fully appreciated by conventional SUV-based metrics.

physics.med-ph↗

arXiv2025

Performance Evaluation of a Position-Sensitive SiPM-based Gamma Camera for Intraoperative Imaging

The POSiCS camera is a handheld, small field-of-view gamma camera developed for multipurpose use in radio-guided surgery (RGS), with sentinel lymph node biopsy (SLNB) as its benchmark application. This compact and lightweight detector (weighing approximately 350 g) can map tissues labeled with Tc-99m nanocolloids and guide surgeons to the location of target lesions. By enabling intraoperative visualization in close proximity to the surgical field, its primary objective is to minimize surgical interventional invasiveness and operative time, thereby enhancing localization accuracy and reducing the incidence of post-operative complications. The design and components of the POSiCS camera emphasize ergonomic handling and compactness, providing, at the same time, rapid image formation and a spatial resolution of a few millimeters. These features are compatible with routine operating-room workflow, including wireless communication with the computer and a real-time display to support surgeon decision-making. The spatial resolution measured at a source-detector distance of 0 cm was 1.9 +/- 0.1 mm for the high-sensitivity mode and 1.4 +/- 0.1 mm for the high-resolution mode. The system sensitivity at 2 cm was evaluated as 481 +/- 14 cps/MBq (high sensitivity) and 134 +/- 8 cps/MBq (high resolution). For both working modes, we report an energy resolution of approximately 20 percent, even though the high-resolution collimator exhibits an increased scattered component due to the larger amount of tungsten.

physics.med-ph↗

arXiv2025

Mitigating MRI Domain Shift in Sex Classification: A Deep Learning Approach with ComBat Harmonization

Deep learning models for medical image analysis often suffer from performance degradation when applied to data from different scanners or protocols, a phenomenon known as domain shift. This study investigates this challenge in the context of sex classification from 3D T1-weighted brain magnetic resonance imaging (MRI) scans using the IXI and OASIS3 datasets. While models achieved high within-domain accuracy (around 0.95) when trained and tested on a single dataset (IXI or OASIS3), we demonstrate a significant performance drop to chance level (about 0.50) when models trained on one dataset are tested on the other, highlighting the presence of a strong domain shift. To address this, we employed the ComBat harmonization technique to align the feature distributions of the two datasets. We evaluated three state-of-the-art 3D deep learning architectures (3D ResNet18, 3D DenseNet, and 3D EfficientNet) across multiple training strategies. Our results show that ComBat harmonization effectively reduces the domain shift, leading to a substantial improvement in cross-domain classification performance. For instance, the cross-domain balanced accuracy of our best model (ResNet18 3D with Attention) improved from approximately 0.50 (chance level) to 0.61 after harmonization. t-SNE visualization of extracted features provides clear qualitative evidence of the reduced domain discrepancy post-harmonization. This work underscores the critical importance of domain adaptation techniques for building robust and generalizable neuroimaging AI models.

physics.med-ph↗

arXiv2025

Attention-Enhanced Deep Learning Ensemble for Breast Density Classification in Mammography

Breast density assessment is a crucial component of mammographic interpretation, with high breast density (BI-RADS categories C and D) representing both a significant risk factor for developing breast cancer and a technical challenge for tumor detection. This study proposes an automated deep learning system for robust binary classification of breast density (low: A/B vs. high: C/D) using the VinDr-Mammo dataset. We implemented and compared four advanced convolutional neural networks: ResNet18, ResNet50, EfficientNet-B0, and DenseNet121, each enhanced with channel attention mechanisms. To address the inherent class imbalance, we developed a novel Combined Focal Label Smoothing Loss function that integrates focal loss, label smoothing, and class-balanced weighting. Our preprocessing pipeline incorporated advanced techniques, including contrast-limited adaptive histogram equalization (CLAHE) and comprehensive data augmentation. The individual models were combined through an optimized ensemble voting approach, achieving superior performance (AUC: 0.963, F1-score: 0.952) compared to any single model. This system demonstrates significant potential to standardize density assessments in clinical practice, potentially improving screening efficiency and early cancer detection rates while reducing inter-observer variability among radiologists.

eess.IV↗

arXiv2025

Quantitative Determination of Spatial Resolution and Linearity of Position-Sensitive LG-SiPMs at Sub-Millimeter Scale via Ricean Distribution Fitting

Position-sensitive SiPMs are useful in all light detection applications requiring a small number of readout channels while preserving the information about the incoming light's interaction position. Focusing on a 2x2 array of LG-SiPMs covering an area of $\sim 15.5 \times 15.5~\rm{mm}$ with just 6 readout channels, we proposed a quantitative method to evaluate image reconstruction performance. The method is based on a statistical approach to assess the device's precision (spatial resolution) and accuracy (linearity) in reconstructing the light spot center of gravity. This evaluation is achieved through a Rice probability distribution function fitting. We obtained an average sensor spatial resolution's best value of $81 \pm 3~\rm{\mu m}$ (standard deviation), which is achieved by reconstructing each position with the amplitude of the channels' output signals. The corresponding accuracy is $231 \pm 4~\rm{\mu m}$.

physics.ins-det↗

arXiv2022

A hierarchical approach for pulmonary nodules identification from CT images using YOLO v5s nodule detection and 3D neural network classifier

In the first step, a pre-trained model (YOLO) was used to detect all suspicious nod-ules. The YOLO model was re-trained using 397 CT images to detect the entire nodule in CT images. To maximize the sensitivity of the model, a confidence level (the probability threshold for object detection) of 0.3 was set for nodule detection in the first phase (ensuring the entire suspicious nodules are detected from the input CT images). The aim of the hierarchy model is to detect and classify the entire lung nodules (from CT images) with a low false-negative rate. Given the outcome of the first step, we proposed a 3D CNN classifier to analyze and classify the suspicious nodules detected by the YOLO model to achieve a nodule detection framework with a very low false-negative rate. This framework was evaluated using the LUNA 16 dataset, which consists of 888 CT images containing the location of 1186 nodules and 400000 non-nodules in the lung. Results: A large number of false positives were detected due to the low confidence level used in the YOLO model. Utilizing the 3D classifier, the accuracy of nodule detection was remarkably enhanced. The YOLO model detected 294 suspicious nodules (out of 321) when using a confidence level of 50%, wherein there were 107 false positives (187 true positives). By reducing the confidence level to 30%, 459 suspicious nodules were identified by the YOLO model, wherein 138 were false positives, and 321 were true positives. When the outcome of the YOLO model with a confidence level of 30% was fed into the 3D CNN classifier, a nodule detection accuracy of 98.4% and AUC of 98.9% were achieved Conclusion: The proposed framework resulted in a few false-negative and false-positives pre-dictions in nodule detection from CT images. The proposed approach would be helpful in detecting pulmonary nodules from CT images as a decision support tool.

physics.med-ph↗

arXiv2022

Deep Learning-Based Partial Volume Correction in Standard and Low-Dose PET-CT Imaging

A standard dose of radioactive tracer must be delivered into the patients body to obtain high-quality Positron Emission Tomography (PET) images for diagnostic purposes, which raises the risk of radiation harm. A reduced tracer dose, on the other hand, results in poor image quality and a noise-induced quantitative bias in PET imaging. The partial volume effect (PVE), which is the result of PET intrinsic limited spatial resolution, is another source of quality and quantity degradation in PET imaging. The utilization of anatomical information for PVE correction (PVC) is not straightforward due to the internal organ motions, patient involuntary motions, and discrepancies in the appearance and size of the structures in anatomical and functional images. Furthermore, an additional MR imaging session is necessary for anatomical information, which may not be available. We set out to build a deep learning-based framework for predicting partial volume corrected full-dose (FD-PVC) pictures from either standard or low-dose (LD) PET images without requiring any anatomical data in order to provide a joint solution for PVC and denoise low-dose PET images.

physics.med-ph↗

arXiv2022

Deep Learning-Based Attenuation and Scatter Correction of Brain 18F-FDG PET Images in the Image Domain

Attenuation and scatter correction (AC) is crucial for quantitative Positron Emission Tomography (PET) imaging. Recently, direct application of AC in the image domain using deep learning approaches has been proposed for the hybrid PET/MR and dedicated PET systems that lack accompanying transmission or anatomical imaging. This study set out to investigate deep learning-based AC in the image domain using different input settings.

physics.med-ph↗

arXiv2022

Design and implementation of brain surgery bipolar electrocautery simulator using haptic technology

Surgical simulators have been widely used in training and evaluation of physicians and surgeons. Virtual reality augmented with haptic technology has made it feasible to develop more realistic surgical simulators. In this context, we set out to design and develop a brain surgery bipolar electrocautery simulator using haptic technology. A 3D model of brain tissue was generated based on a brain craniotomy image. Bipolar forceps were also modeled to visually assimilate real forceps. An experiment was developed to assess the learning process of the participants. In this experiment, the volunteers were asked to cauterize a large blood vessel in the brain while minimizing the damage done to the brain tissue. The experiment was performed on 20 volunteers, and statistical analysis was conducted on the learning process and error reduction during the surgery. Next, the volunteers were divided into gamer and non gamer groups. The analysis of the volunteers operation demonstrated that, on average, there was a 5 percent reduction in the percentage of applied force error. It was also shown that the results achieved by the gamer and non gamer group has significant difference with a p value of 0.0001. So, playing computer games would increase hand control, focus, and reflex and positively affect surgery skills.

physics.med-ph↗

arXiv2022

A novel shape-based loss function for machine learning-based seminal organ segmentation in medical imaging

Automated medical image segmentation is an essential task to aid/speed up diagnosis and treatment procedures in clinical practices. Deep convolutional neural networks have exhibited promising performance in accurate and automatic seminal segmentation. For segmentation tasks, these methods normally rely on minimizing a cost/loss function that is designed to maximize the overlap between the estimated target and the ground-truth mask delineated by the experts. A simple loss function based on the degrees of overlap (i.e., Dice metric) would not take into account the underlying shape and morphology of the target subject, as well as its realistic/natural variations; therefore, suboptimal segmentation results would be observed in the form of islands of voxels, holes, and unrealistic shapes or deformations. In this light, many studies have been conducted to refine/post-process the segmentation outcome and consider an initial guess as prior knowledge to avoid outliers and/or unrealistic estimations. In this study, a novel shape-based cost function is proposed which encourages/constrains the network to learn/capture the underlying shape features in order to generate a valid/realistic estimation of the target structure. To this end, the Principal Component Analysis (PCA) was performed on a vectorized training dataset to extract eigenvalues and eigenvectors of the target subjects. The key idea was to use the reconstruction weights to discriminate valid outcomes from outliers/erroneous estimations.

physics.med-ph↗

arXiv2022

Joint brain tumor segmentation from multi MR sequences through a deep convolutional neural network

Brain tumor segmentation is highly contributive in diagnosing and treatment planning. The manual brain tumor delineation is a time-consuming and tedious task and varies depending on the radiologists skill. Automated brain tumor segmentation is of high importance, and does not depend on either inter or intra-observation. The objective of this study is to automate the delineation of brain tumors from the FLAIR, T1 weighted, T2 weighted, and T1 weighted contrast-enhanced MR sequences through a deep learning approach, with a focus on determining which MR sequence alone or which combination thereof would lead to the highest accuracy therein.

physics.med-ph↗

arXiv2022

A novel unsupervised covid lung lesion segmentation based on the lung tissue identification

This study aimed to evaluate the performance of a novel unsupervised deep learning-based framework for automated infections lesion segmentation from CT images of Covid patients. In the first step, two residual networks were independently trained to identify the lung tissue for normal and Covid patients in a supervised manner. These two models, referred to as DL-Covid and DL-Norm for Covid-19 and normal patients, respectively, generate the voxel-wise probability maps for lung tissue identification. To detect Covid lesions, the CT image of the Covid patient is processed by the DL-Covid and DL-Norm models to obtain two lung probability maps. Since the DL-Norm model is not familiar with Covid infections within the lung, this model would assign lower probabilities to the lesions than the DL-Covid. Hence, the probability maps of the Covid infections could be generated through the subtraction of the two lung probability maps obtained from the DL-Covid and DL-Norm models. Manual lesion segmentation of 50 Covid-19 CT images was used to assess the accuracy of the unsupervised lesion segmentation approach. The Dice coefficients of 0.985 and 0.978 were achieved for the lung segmentation of normal and Covid patients in the external validation dataset, respectively. Quantitative results of infection segmentation by the proposed unsupervised method showed the Dice coefficient and Jaccard index of 0.67 and 0.60, respectively. Quantitative evaluation of the proposed unsupervised approach for Covid-19 infectious lesion segmentation showed relatively satisfactory results. Since this framework does not require any annotated dataset, it could be used to generate very large training samples for the supervised machine learning algorithms dedicated to noisy and/or weakly annotated datasets.

eess.IV↗

arXiv2022

Does prior knowledge in the form of multiple low-dose PET images (at different dose levels) improve standard-dose PET prediction?

Reducing the injected dose would result in quality degradation and loss of information in PET imaging. To address this issue, deep learning methods have been introduced to predict standard PET images (S-PET) from the corresponding low-dose versions (L-PET). The existing deep learning-based denoising methods solely rely on a single dose level of PET images to predict the S-PET images. In this work, we proposed to exploit the prior knowledge in the form of multiple low-dose levels of PET images (in addition to the target low-dose level) to estimate the S-PET images.

physics.med-ph↗

arXiv2021

Characterization of Fe_3 O_4/Au-Ag@MoS_2 nanoparticles for brain cancer treatment using magneto plasmonic approach

This study investigates the treatment of brain cancer by the magnetic hyperthermia approach and nanoparticles including Fe_3 O_4 core with gold, silver alloy shell, and MoS_2 coating. Optical properties of these nanoparticles within the tumor, including the extinction coefficient and surface plasmon peak (SPR) as a function of size, structure, different compositions, and thickness are also investigated using the effective medium theory. Moreover, the impact of temperature distribution was assessed through the analytical modeling of alternating current (AC) magnetic field. The results of this study indicated that nanoparticles with a compound of Fe_3 O_4 - Au_0.25 Ag_0.75@MoS_2 and a thickness of 3 nm of gold-silver alloy and 3 layers of MoS_2 have the best coefficient of extinction and SPR in the biological window. The gold-silver alloy improved the extinction coefficient and, at the same time, prevented the accumulation of magnetic nanoparticles. Since the gold-silver alloy alone cannot function within the range of biological windows, MoS_2 was used, which increased the extinction efficiency at higher wavelengths. Examination of the temperature distribution in the tumor for the proposed alloy compound indicated that after a short time from the start of irradiation, the tumor temperature reaches 45 C degree. Also, the temperature distribution within the tumor tissue reached its maximum value at the center of the tumor and decreased dramatically as getting away from the center. The use of magnetic hyperthermia enabled localized delivery of therapeutic dose to malignant brain tumors; hence, exhibiting superior performance/efficiency over the photothermal method.

physics.med-ph↗

arXiv2021

The impact of MR-guided attenuation correction (compared to CTbased AC) on the diagnosis of anosmia based on 99m-Tc EthylCysteinate-Dimer SPECT images

99m-Tc Ethyl-Cysteinate-Dimer SPECT and MR imaging play a significant role in diagnosing anosmia. In this study, two-tissue class and three-tissue class attenuation maps (2C-MR and 3CMR) obtained from MR images were compared with CT-based attenuation correction (CTAC). Afterward, the presence of hypo-perfusion in brain lobes was evaluated in SPECT images. The 2C-MRAC map was generated through segmentation of T1-W MR images into air and soft-tissue, while in the 3C-MRAC map, the cortical bone was also considered. For investigating MRAC approaches, the difference between activity concentration (ACC) values was estimated in 144 volumes of interest. Ten normal and fourteen anosmic patients were compared by calculating the average normalized count and standard uptake value ratio parameters in the brain lobes. The comparison between attenuation correction strategies represented that MRAC images resulted in underestimation of the ACC values which was more substantial in the cortical area rather than in central regions (maximum 9% vs. 6% for 2C-MR and maximum 5.5% vs. 3.5% for 3C-MR). Nevertheless, there was a strong correlation between the MRAC and CTAC methods with a correlation coefficient of 0.7 for both 2C-MR and 3C-MR. The statistical analysis between normal and affected groups indicated the hypoperfusion in the cortex of Lh_frontal, Rh and Lh_temporal lobes with p-values < 0.05. Using MRAC resulted in underestimation of activity concentration which was partly eliminated by considering the cortical bone in the 3C-MR attenuation map. Hypo-perfusion was perceived in Frontal and Temporal lobes in SPECT-MRAC images of the anosmic group.

physics.med-ph↗

arXiv2021

Quantitative and Qualitative Performance Evaluation of Commercial Metal Artifact Reduction Methods: Dosimetric Effects on the Treatment Planning

The presence of metal implants within CT imaging causes severe attenuation of the X-ray beam. Due to the incomplete information recorded by CT detectors, artifacts in the form of streaks and dark bands would appear in the resulting CT images. The metal-induced artifacts would firstly affect the quantitative accuracy of CT imaging, and consequently, the radiation treatment planning and dose estimation in radiation therapy. To address this issue, CT scanner vendors have implemented metal artifact reduction (MAR) algorithms to avoid such artifacts and enhance the overall quality of CT images. The orthopedic-MAR (OMAR) and normalized MAR (NMAR) algorithms are the most well-known metal artifact reduction (MAR) algorithms, used worldwide. These algorithms have been implemented on Philips and Siemens scanners, respectively. In this study, we set out to quantitatively and qualitatively evaluate the effectiveness of these two MAR algorithms and their impact on accurate radiation treatment planning and CT-based dosimetry. The quantitative metrics measured on the simulated metal artifact dataset demonstrated superior performance of the OMAR technique over the NMAR one in metal artifact reduction. The analysis of radiation treatment planning using the OMAR and NMAR techniques in the corrected CT images showed that the OMAR technique reduced the toxicity of healthy tissues by 10% compared to the uncorrected CT images.

physics.med-ph↗