Machine Learning Applied to Pre-Operative Computed-Tomography-Based Radiomic Features Can Accurately Differentiate Uterine Leiomyoma from Leiomyosarcoma: A Pilot Study.

BACKGROUND: The accurate discrimination of uterine leiomyosarcomas and leiomyomas in a pre-operative setting remains a current challenge. To date, the diagnosis is made by a pathologist on the excised tumor. The aim of this study was to develop a machine learning algorithm using radiomic data extracted from contrast-enhanced computed tomography (CECT) images that could accurately distinguish leiomyosarcomas from leiomyomas. METHODS: Pre-operative CECT images from patients submitted to surgery with a histological diagnosis of leiomyoma or leiomyosarcoma were used for the region of interest identification and radiomic feature extraction. Feature extraction was conducted using the PyRadiomics library, and three feature selection methods combined with the general linear model (GLM), random forest (RF), and support vector machine (SVM) classifiers were built, trained, and tested for the binary classification task (malignant vs. benign). In parallel, radiologists assessed the diagnosis with or without clinical data. RESULTS: A total of 30 patients with leiomyosarcoma (mean age 59 years) and 35 patients with leiomyoma (mean age 48 years) were included in the study, comprising 30 and 51 lesions, respectively. Out of nine machine learning models, the three feature selection methods combined with the GLM and RF classifiers showed good performances, with predicted area under the curve (AUC), sensitivity, and specificity ranging from 0.78 to 0.97, from 0.78 to 1.00, and from 0.67 to 0.93, respectively, when compared to the results obtained from experienced radiologists when blinded to the clinical profile (AUC = 0.73 95%CI = 0.62-0.84), as well as when the clinical data were consulted (AUC = 0.75 95%CI = 0.65-0.85). CONCLUSIONS: CECT images integrated with radiomics have great potential in differentiating uterine leiomyomas from leiomyosarcomas. Such a tool can be used to mitigate the risks of eventual surgical spread in the case of leiomyosarcoma and allow for safer fertility-sparing treatment in patients with benign uterine lesions.

A single centre intercomparison between commercial treatment planning systems for 90Y radioembolization using virtual and experimental phantoms.

INTRODUCTION: Dedicated Treatment Planning Systems (TPSs) were developed to personalize 90Y-transarterial radioembolization. This study evaluated the agreement among four commercial TPSs assessing volumes of interest (VOIs) volumes and dose metrics. METHODS: A homogeneous (EH) and an anthropomorphic phantom with hot and cold inserts (EA) filled with 99mTc-pertechnetate were acquired with a SPECT/CT scanner. Their virtual versions (VH and VA, respectively) and a phantom with activity inside a single voxel (VK) were generated by an in-house MATLAB script. Images and delineated VOIs were imported into the TPSs to compute voxel-based absorbed dose distributions with various dose deposition approaches: local deposition method (LDM) and dose kernel convolution (DKC) with/without local density correction (LDC). VOI volumes and mean absorbed doses were assessed against their median value across TPSs. Dose-volume histograms (DVHs) and VK-derived dose profiles were evaluated. RESULTS: Small (<2.1 %) and large (up to 42.4 %) relative volume differences were observed on large (>500 ml) and small VOIs, respectively. Mean absorbed doses relative differences were < 3 % except for small VOIs with steep dose gradients (up to 89.1 % in the VA Cold Sphere VOI). Within the same TPS, LDC negligibly affected the mean absorbed dose, while DKC and LDM showed differences up to 63 %. DHVs were mostly overlapped in experimental phantoms, with some differences in the virtual versions. Dose profiles agreed within 1 %. CONCLUSION: TPSs showed an overall good agreement except for small VOI volumes and mean absorbed doses of VOIs with steep dose gradients. These discrepancies should be considered in the dosimetry uncertainty assessment, thus requiring an appropriate harmonization.

A Radiomic-Based Machine Learning Model Predicts Endometrial Cancer Recurrence Using Preoperative CT Radiomic Features: A Pilot Study.

BACKGROUND: Current prognostic models lack the use of pre-operative CT images to predict recurrence in endometrial cancer (EC) patients. Our study aimed to investigate the potential of radiomic features extracted from pre-surgical CT scans to accurately predict disease-free survival (DFS) among EC patients. METHODS: Contrast-Enhanced CT (CE-CT) scans from 81 EC cases were used to extract the radiomic features from semi-automatically contoured volumes of interest. We employed a 10-fold cross-validation approach with a 6:4 training to test set and utilized data augmentation and balancing techniques. Univariate analysis was applied for feature reduction leading to the development of three distinct machine learning (ML) models for the prediction of DFS: LASSO-Cox, CoxBoost and Random Forest (RFsrc). RESULTS: In the training set, the ML models demonstrated AUCs ranging from 0.92 to 0.93, sensitivities from 0.96 to 1.00 and specificities from 0.77 to 0.89. In the test set, AUCs ranged from 0.86 to 0.90, sensitivities from 0.89 to 1.00 and specificities from 0.73 to 0.90. Patients classified as having a high recurrence risk prediction by ML models exhibited significantly worse DSF (p-value < 0.001) across all models. CONCLUSIONS: Our findings demonstrate the potential of radiomics in predicting EC recurrence. While further validation studies are needed, our results underscore the promising role of radiomics in forecasting EC outcomes.

How smart is artificial intelligence in organs delineation? Testing a CE and FDA-approved Deep-Learning tool using multiple expert contours delineated on planning CT images.

Background: A CE- and FDA-approved cloud-based Deep learning (DL)-tool for automatic organs at risk (OARs) and clinical target volumes segmentation on computer tomography images is available. Before its implementation in the clinical practice, an independent external validation was conducted. Methods: At least a senior and two in training Radiation Oncologists (ROs) manually contoured the volumes of interest (VOIs) for 6 tumoral sites. The auto-segmented contours were retrieved from the DL-tool and, if needed, manually corrected by ROs. The level of ROs satisfaction and the duration of contouring were registered. Relative volume differences, similarity indices, satisfactory grades, and time saved were analyzed using a semi-automatic tool. Results: Seven thousand seven hundred sixty-five VOIs were delineated on the CT images of 111 representative patients. The median (range) time for manual VOIs delineation, DL-based segmentation, and subsequent manual corrections were 25.0 (8.0-115.0), 2.3 (1.2-8) and 10.0 minutes (0.3-46.3), respectively. The overall time for VOIs retrieving and modification was statistically significantly lower than for manual contouring (p<0.001). The DL-tool was generally appreciated by ROs, with 44% of vote 4 (well done) and 43% of vote 5 (very well done), correlated with the saved time (p<0.001). The relative volume differences and similarity indexes suggested a better inter-agreement of manually adjusted DL-based VOIs than manually segmented ones. Conclusions: The application of the DL-tool resulted satisfactory, especially in complex delineation cases, improving the ROs inter-agreement of delineated VOIs and saving time.

A novel tool for predicting the dose distribution of non-sealed 188 Re (Rhenium) resin in non-melanoma skin cancers (NMSC) patients.

BACKGROUND: High-dose rate brachytherapy using a non-sealed 188 Rhenium resin (188 Re) is a recently approved treatment option for non-melanoma skin cancer (NMSC). The treatment goal is to deliver a personalized absorbed dose to the deepest point of neoplastic infiltration corresponding to the minimal target dose. The treatment consists of the application of a 188 Re-based resin over a plastic foil placed on the target skin surface. However, there is no treatment planning tool to assess the 188 Re activity needed for a personalized treatment. PURPOSE: The paper aims to present a novel Monte Carlo (MC)-based tool for 188 Re-based resin activity and dose calculation, experimentally validated using Gafchromic EBT3 films. METHODS: MC simulations were carried out using FLUKA modeling density and composition of 188 Re resin. The MC-based look up table (LUT) was incorporated in an ad hoc developed tool. The proposed tool allows the personalized calculation of treatment parameters (i.e., activity to be dispensed, the treatment duration, and dose volume histograms), according to the target dimension. The proposed tool was compared using Bland-Altman analysis to the previous calculation approaches conducted using VARSKIN in a retrospective cohort of 76 patients. The tool was validated in ad hoc experimental set ups using a stack of calibrated Gafchromic EBT3 films covered by a plastic film and exposed using a homogenous activity distribution of 188 Re eluate and a heterogeneous activity distribution of 188 Re resin mimic the patient treatment. RESULTS: The agreement between the proposed tool and VARSKIN was evaluated on the investigated cohort with median range of target area, target depth, and treatment time equal to 4.8 [1.0-60.1] cm2 , 1.1 [0.2-3.0] mm, and 70 [21-285] min, with a median range of target dose (Gy) of 23.5 [10-54.9]. The calculated minimal target doses, ranged from 1% to 10% for intermediate target depths (1.2 ± 0.7 mm), while showing significant differences in the estimation of superficial (maximal) target doses. The agreement between MC calculation and measurements at different plans in a stack of Gafchromic EBT3 films was within 10% for both the homogenous and heterogeneous activity distribution of 188 Re. Worst agreements were observed for absorbed doses lower than 0.3 Gy. CONCLUSIONS: Our results support the implementation of our MC-based tool in the practical routine for calculating the 188 Re resin activity and treatment parameters necessary for obtaining the prescribed minimal target dose.

Using Cone Beam Computed Tomography for Radiological Assessment Beyond Dento-maxillofacial Imaging: A Review of the Clinical Applications in other Anatomical Districts.

BACKGROUND: Cone Beam Computed Tomography (CBCT) represents the optimal imaging solution for the evaluation of the maxillofacial and dental area when quantitative geometric and volumetric accuracy is necessary (e.g., in implantology and orthodontics). Moreover, in recent years, this technique has given excellent results for the imaging of lower and upper extremities. Therefore, significant interest has been increased in using CBCT to investigate larger and non-traditional anatomical districts. OBJECTIVE: The purpose of this work is to review the scientific literature in Pubmed and Scopus on CBCT application beyond head districts by paying attention to image quality and radiological doses. METHODS: The search for keywords was conducted in Pubmed and Scopus databases with no back-date restriction. Papers on applications of CBCT to head were excluded from the present work. From each considered paper, parameters related to image quality and radiological dose were extracted. An overall qualitative evaluation of the results extracted from each issue was done by comparing the conclusive remarks of each author regarding doses and image quality. PRISMA statements were followed during this process. RESULTS: The review retrieved 97 issues from 83 extracted papers; 46 issues presented a comparison between CBCT and Multi-Detector Computed Tomography (MDCT), and 51 reviewed only CBCT. The radiological doses given to the patient with CBCT were considered acceptable in 91% of cases, and the final image quality was found in 99%. CONCLUSION: CBCT represents a promising technology not only for imaging of the head and upper and lower extremities but for all the orthopedic districts. Moreover, the application of CBCT derived from C-arms (without the possibility of a 360 ° rotation range) during invasive investigations demonstrates the feasibility of this technique for non-standard anatomical areas, from soft tissues to vascular beds, despite the limits due to the incomplete rotation of the tube.

A novel figure of merit to investigate 68Ga PET/CT image quality based on patient weight and lesion size using Q.Clear reconstruction algorithm: A phantom study.

INTRODUCTION: Q.Clear is a Bayesian penalised-likelihood algorithm that uses a ß-value for positron emission tomography(PET)/computed tomography(CT) image reconstruction(IR). Our study proposes a novel figure of merit, named CRBV, to compare the Q.Clear performances using 68Ga PET/CT image with the ordered-subset-expectation-maximization(OSEM) algorithm and to identify the optimal ß-values for these images using two phantoms mimicking normal and overweight patients. METHODS: NEMA IQ phantom with or without a ring of water-filled plastic bags (NEMAstd and NEMAow, respectively) was acquired and reconstructed with OSEM and Q.Clear at various ß-values and minutes/bed position(min/bp). Contrast recovery(CR), background variability(BV) and CRBV were calculated. Highest CRBV values were used to identify optimal ß-value ranges. RESULTS: Q.Clear with 250 ≤ ß ≤ 800 improved CRBV compared to OSEM for all the investigated spheres and acquisition setups. Outside of this range, Q.Clear still outperformed OSEM with few exceptions depending on spheres diameters and phantoms(e.g.,ß-value = 1600 for diameters ≤ 17 mm using the NEMAow phantom). Regarding the CRBV performance for IR optimization, for the 4 min/bp NEMAstd IR, ß-values = 300 ÷ 350 allowed to simultaneously optimize all diameters(except for the 10 mm); for the NEMAow IR, ß-values = 350 ÷ 500 were needed for diameters > 20 mm, while ß-values = 200 ÷ 250 were selected for the remaining diameters. For the 2 min/bp, ß-value = 500 was suitable for diameters > 17 mm in both NEMAstd and NEMAow IR, while for smaller diameters ß-value = 200 and ß-values = 250 ÷ 350 were obtained for NEMAstd and NEMAow, respectively. CONCLUSION: Almost all tested ß-values of Q.Clear improved the CRBV compared to OSEM. In both phantoms, simulating normal and over-weight patients, optimal ß-values were found according to lesion sizes and investigated acquisition times.

PSMA PET for the Evaluation of Liver Metastases in Castration-Resistant Prostate Cancer Patients: A Multicenter Retrospective Study.

Background: To evaluate the diagnostic performance of PSMA-PET compared to conventional imaging/liver biopsy in the detection of liver metastases in CRPC patients. Moreover, we evaluated a PSMA-PET/CT-based radiomic model able to identify liver metastases. Methods: Multicenter retrospective study enrolling patients with the following inclusion criteria: (a) proven CRPC patients, (b) PSMA-PET and conventional imaging/liver biopsy performed in a 6 months timeframe, (c) no therapy changes between PSMA-PET and conventional imaging/liver biopsy. PSMA-PET sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy for liver metastases were calculated. After the extraction of radiomic features, a prediction model for liver metastases identification was developed. Results: Sixty CRPC patients were enrolled. Within 6 months before or after PSMA-PET, conventional imaging and liver biopsy identified 24/60 (40%) patients with liver metastases. PSMA-PET sensitivity, specificity, PPV, NPV, and accuracy for liver metastases were 0.58, 0.92, 0.82, 0.77, and 0.78, respectively. Either number of liver metastases and the maximum lesion diameter were significantly associated with the presence of a positive PSMA-PET (p < 0.05). On multivariate regression analysis, the radiomic feature-based model combining sphericity, and the moment of inverse difference (Idm), had an AUC of 0.807 (95% CI:0.686-0.920). Conclusion: For liver metastases assessment, [68Ga]Ga-PSMA-11-PET demonstrated moderate sensitivity while high specificity, PPV, and inter-reader agreement compared to conventional imaging/liver biopsy in CRPC patients.

A novel tool for motion-related dose inaccuracies reduction in 99mTc-MAA SPECT/CT images for SIRT planning.

INTRODUCTION: In Selective Internal Radiation Therapy (SIRT), 90Y is administered to primary/secondary hepatic lesions. An accurate pre-treatment planning using 99mTc-MAA SPECT/CT allows the assessment of its feasibility and of the activity to be injected. Unfortunately, SPECT/CT suffers from patient-specific respiratory motion which causes artifacts and absorbed dose inaccuracies. In this study, a data-driven solution was developed to correct the respiratory motion. METHODS: The tool realigns the barycenter of SPECT projection images and shifts them to obtain a fine registration with the attenuation map. The tool was validated using a modified dynamic phantom with several breathing patterns. We compared the absorbed dose distributions derived from uncorrected(Dm)/corrected(Dc) images with static ones(Ds) in terms of γ-passing rates, 210 Gy isodose volumes, dose-volume histograms and percentage differences of mean doses (i.e., ΔD¯m and ΔD¯c, respectively). The tool was applied to twelve SIRT patients and the Bland-Altman analysis was performed on mean doses. RESULTS: In the phantom study, the agreement between Dc and Ds was higher (γ-passing rates generally > 90%) than Dm and Ds. The isodose volumes in Dc were closer than Dm to Ds, with differences up to 10% and 30% respectively. A reduction from a median ΔD¯m = -19.3% to ΔD¯c = -0.9%, from ΔD¯m = -42.8% to ΔD¯c = -7.0% and from ΔD¯m = 1586% to ΔD¯c = 47.2% was observed in liver-, tumor- and lungs-like structures. The Bland-Altman analysis on patients showed variations (±50 Gy) and (±4 Gy) between D¯c and D¯m of tumor and lungs, respectively. CONCLUSION: The proposed tool allowed the correction of 99mTc-MAA SPECT/CT images, improving the accuracy of the absorbed dose distribution.

Improving total body irradiation with a dedicated couch and 3D-printed patient-specific lung blocks: A feasibility study.

Introduction: Total body irradiation (TBI) is an important component of the conditioning regimen in patients undergoing hematopoietic stem cell transplants. TBI is used in very few patients and therefore it is generally delivered with standard linear accelerators (LINACs) and not with dedicated devices. Severe pulmonary toxicity is the most common adverse effect after TBI, and patient-specific lead blocks are used to reduce mean lung dose. In this context, online treatment setup is crucial to achieve precise positioning of the lung blocks. Therefore, in this study we aim to report our experience at generating 3D-printed patient-specific lung blocks and coupling a dedicated couch (with an integrated onboard image device) with a modern LINAC for TBI treatment. Material and methods: TBI was planned and delivered (2Gy/fraction given twice a day, over 3 days) to 15 patients. Online images, to be compared with planned digitally reconstructed radiographies, were acquired with the couch-dedicated Electronic Portal Imaging Device (EPID) panel and imported in the iView software using a homemade Graphical User Interface (GUI). In vivo dosimetry, using Metal-Oxide Field-Effect Transistors (MOSFETs), was used to assess the setup reproducibility in both supine and prone positions. Results: 3D printing of lung blocks was feasible for all planned patients using a stereolithography 3D printer with a build volume of 14.5×14.5×17.5 cm3. The number of required pre-TBI EPID-images generally decreases after the first fraction. In patient-specific quality assurance, the difference between measured and calculated dose was generally<2%. The MOSFET measurements reproducibility along each treatment and patient was 2.7%, in average. Conclusion: The TBI technique was successfully implemented, demonstrating that our approach is feasible, flexible, and cost-effective. The use of 3D-printed patient-specific lung blocks have the potential to personalize TBI treatment and to refine the shape of the blocks before delivery, making them extremely versatile.

Dual-Energy Computed Tomography Applications to Reduce Metal Artifacts in Hip Prostheses: A Phantom Study.

Metal components of hip prostheses cause severe artifacts in CT images, influencing diagnostic accuracy. Metal artifact reduction (MAR) software and virtual monoenergetic reconstructions on dual-energy CT (DECT) systems are possible solutions that should be considered. In this study, we created a customized adjustable phantom to quantify the severity of artifacts on periprosthetic tissues (cortical and spongious bone, soft tissues) for hip prostheses. The severity of artifacts was classified by different thresholds of deviation from the CT numbers for reference objects not affected by artifacts. The in vitro setup was applied on four unilateral and three bilateral configurations of hip prostheses (made of titanium, cobalt, and stainless steel alloys) with a DECT system, changing the energy of virtual monoenergetic reconstructions, with and without MAR. The impact of these tools on the severity of artifacts was scored, looking for the best scan conditions for the different configurations. For titanium prostheses, the reconstruction at 110 keV, without MAR, always minimized the artifacts. For cobalt and stainless-steel prostheses, MAR should always be applied, while monoenergetic reconstruction alone did not show clear advantages. The available tools for reducing metal artifacts must therefore be applied depending on the examined prosthetic configuration.

A novel tool for assessing the correlation of internal/external markers during SGRT guided stereotactic ablative radiotherapy treatments.

INTRODUCTION: An in-house developed tool was implemented and validated to investigate the skin surface, hepatic dome, and target displacement for stereotactic ablative radiotherapy (SABR) of thoracic/abdominal lesions using a Surface Guided Radiation Therapy (SGRT) system combined with 4D- images. MATERIALS AND METHODS: Fourteen consecutive patients with tumors near the hepatic dome undergoing SABR treatments were analyzed. For each patient, a planning 4D-CT and five 4D-CBCT images were acquired. The C-RAD technology was also used to register/monitor the position of the skin reference point (SRP) as an external marker representative of patient breathing. The 4D images were imported in the developed tool, and the absolute maximum height (Pmax,dome) of the hepatic dome on the ten respiratory phases was semi-automatically detected. Similarly, the contour of the skin surface was extracted in correspondence with the SRP position. The tool has been validated using an ad hoc modified moving phantom with pre-selected amplitudes and numbers of cycles. The Pearson correlation coefficients and Bland-Altman plots were calculated. RESULTS: There was a strong correlation between the skin motion amplitude based on 4D-CBCT and the C-RAD in all the patients (0.90 ± 0.08). Similarly, the mean ± SD of Pearson correlation coefficients of skin and Pmax,dome movements registered by 4D-CT and 4D-CBCT were 0.90 ± 0.05 and 0.94 ± 0.05, respectively. The mean ± SD of Pearson correlation coefficients comparing the skin and Pmax,dome displacements within each imaging modality were 0.88 ± 0.05 and 0.90 ± 0.05 for 4D-CT and 4D-CBCT, respectively. The SRP displacement during the set-up imaging and the treatment delivery were similar in all the investigated patients. Similar results were obtained for the ad hoc modified phantom in the preliminary validation phase. CONCLUSION: The strong correlation between the tumor/ hepatic dome and skin displacements confirms that the SGRT approach can be considered appropriate for intra- and inter-fraction motion management in SABR therapy.