Background and purpose:In order to optimize the radiotherapy treatment and minimize toxicities, organs-at-risk (OARs) and clinical target volume (CTV) must be segmented. Deep Learning (DL) techniques show significant potential for performing this task effectively. The availability of a large single-institute data sample, combined with additional numerous multi-centric data, makes it possible to develop and validate a reliable CTV segmentation model. Materials and methods:Planning CT data of 1822 patients were available (861 from a single center for training and 961 from 8 centers for validation). A preprocessing step, aimed at standardizing all the images, followed by a 3D-Unet capable of segmenting both right and left CTVs was implemented. The metrics used to evaluate the performance were the Dice similarity coefficient (DSC), the Hausdorff distance (HD), and its 95th percentile variant (HD_95) and the Average Surface Distance (ASD). Results:The segmentation model achieved high performance on the validation set (DSC: 0.90; HD: 20.5 mm; HD_95: 10.0 mm; ASD: 2.1 mm; epoch 298). Furthermore, the model predicted smoother contours than the clinical ones along the cranial-caudal axis in both directions. When applied to internal and external data the same metrics demonstrated an overall agreement and model transferability for all but one (Inst 9) center. Conclusion:. A 3D-Unet for CTV segmentation trained on a large single institute cohort consisting of planning CTs and manual segmentations was built and externally validated, reaching high performance.

PURPOSE:To train and validate KB prediction models by merging a large multi-institutional cohort of whole breast irradiation (WBI) plans using tangential fields. METHODS:Ten institutions (INST1-INST10, 1481 patients) developed their KB-institutional models for left/right WBI (ten models for right and eight models for left). The transferability of models among centers was assessed based on the overlap of the geometric Principal Component (PC1) of each model when applied to other institutions and/or on the presence of significantly different optimization policies. Centers corresponding to transferable models were asked to join the building of two KB-benchmark models for right/left breast. Dose-volume histogram (DVH) prediction bands (lung/heart) were compared against those of the KB-institutional models. RESULTS:All models were transferable except INST6 (right breast) and INST1 (left breast). Planning data from 6 institutions for right breast and 5 institutions for left breast (out of 9 and 7 institutions with transferable models, respectively) were combined, totaling data from 850 patients. Prediction bands on the test cohorts (n = 30/25 right/left) showed a large overlap with bands of each institution model: for the right-breast, the KB-benchmark model predicts slightly lower lung Dmean when compared to KB-institution models, except for INST7. Regarding the left-breast, even greater similarity between KB-benchmark and KB-institution model predictions was found. CONCLUSIONS:Multi-institutional KB-benchmark models for WBI were successfully generated. They may be employed by other users, representing the performances reached in a multi-institutional context of experienced centers. KB-benchmark models can also have significant applications for large-scale automatic plan optimization, QA/audit and tutoring/education purposes.

Purpose: Within a multi-institutional project, we aimed to assess the transferability of knowledge-based (KB) plan prediction models in the case of whole breast irradiation (WBI) for left-side breast irradiation with tangential fields (TF). Methods: Eight institutions set KB models, following previously shared common criteria. Plan prediction performance was tested on 16 new patients (2 pts per centre) extracting dose-volume-histogram (DVH) prediction bands of heart, ipsilateral lung, contralateral lung and breast. The inter-institutional variability was quantified by the standard deviations (SDint) of predicted DVHs and mean-dose (Dmean). The transferability of models, for the heart and the ipsilateral lung, was evaluated by the range of geometric Principal Component (PC1) applicability of a model to test patients of the other 7 institutions. Results: SDint of the DVH was 1.8 % and 1.6 % for the ipsilateral lung and the heart, respectively (20 %-80 % dose range); concerning Dmean, SDint was 0.9 Gy and 0.6 Gy for the ipsilateral lung and the heart, respectively ( <0.2 Gy for contralateral organs). Mean predicted doses ranged between 4.3 and 5.9 Gy for the ipsilateral lung and 1.1 - 2.3 Gy for the heart. PC1 analysis suggested no relevant differences among models, except for one centre showing a systematic larger sparing of the heart, concomitant to a worse PTV coverage, due to high priority in sparing the left anterior descending coronary artery. Conclusions: Results showed high transferability among models and low inter-institutional variability of 2% for plan prediction. These findings encourage the building of benchmark models in the case of TF-WBI.

(1) Background and (2) Methods: In this retrospective, observational, monocentric study, we selected a cohort of eighty-five patients (age range 38-87 years old, 51 men), enrolled between January 2014 and December 2020, with a newly diagnosed renal mass smaller than 4 cm (SRM) that later underwent nephrectomy surgery (partial or total) or tumorectomy with an associated histopatological study of the lesion. The radiomic features (RFs) of eighty-five SRMs were extracted from abdominal CTs bought in the portal venous phase using three different CT scanners. Lesions were manually segmented by an abdominal radiologist. Image analysis was performed with the Pyradiomic library of 3D-Slicer. A total of 108 RFs were included for each volume. A machine learning model based on radiomic features was developed to distinguish between benign and malignant small renal masses. The pipeline included redundant RFs elimination, RFs standardization, dataset balancing, exclusion of non-reproducible RFs, feature selection (FS), model training, model tuning and validation of unseen data. (3) Results: The study population was composed of fifty-one RCCs and thirty-four benign lesions (twenty-five oncocytomas, seven lipid-poor angiomyolipomas and two renal leiomyomas). The final radiomic signature included 10 RFs. The average performance of the model on unseen data was 0.79 ± 0.12 for ROC-AUC, 0.73 ± 0.12 for accuracy, 0.78 ± 0.19 for sensitivity and 0.63 ± 0.15 for specificity. (4) Conclusions: Using a robust pipeline, we found that the developed RFs signature is capable of distinguishing RCCs from benign renal tumors.

Purpose: To design, fabricate and characterize 3D printed, anatomically realistic, compressed breast phantoms for digital mammography (DM) and digital breast tomosynthesis (DBT) x-ray imaging.Materials: We realized 3D printed phantoms simulating healthy breasts, via fused deposition modeling (FDM), with a layer resolution of 0.1 mm and 100% infill density, using a dual extruder printer. The digital models were derived from a public dataset of segmented clinical breast computed tomography scans. Three physical phantoms were printed in polyethylene terephthalate (PET), acrylonitrile-butadiene-styrene (ABS), or in polylactic-acid (PLA) materials, using ABS as a substitute for adipose tissue, and PLA or PET filaments for replicating glandular and skin tissues. 3D printed phantoms were imaged at three clinical centers with DM and DBT scanners, using typical spectra. Anatomical noise of the manufactured phantoms was evaluated via the estimates of the beta parameter both in DM images and in images acquired via a clinical computed tomography (CT) scanner.Results: DM and DBT phantom images showed an inner texture qualitatively similar to the images of a clinical DM or DBT exam, suitably reproducing the glandular structure of their computational phantoms. beta parameters evaluated in DM images of the manufactured phantoms ranged between 2.84 and 3.79; a lower beta was calculated from the CT scan.Conclusions: FDM 3D printed compressed breast phantoms have been fabricated using ABS, PLA and PET filaments. DM and DBT images with clinical x-ray spectra showed realistic textures. These phantoms appear promising for clinical applications in quality assurance, image quality and dosimetry assessments.

The ongoing COVID-19 pandemic currently involves millions of people worldwide. Radiology plays an important role in the diagnosis and management of patients, and chest computed tomography (CT) is the most widely used imaging modality. An automatic method to characterize the lungs of COVID-19 patients based on individually optimized Hounsfield unit (HU) thresholds was developed and implemented. Lungs were considered as composed of three components—aerated, intermediate, and consolidated. Three methods based on analytic fit (Gaussian) and maximum gradient search (using polynomial and original data fits) were implemented. The methods were applied to a population of 166 patients scanned during the first wave of the pandemic. Preliminarily, the impact of the inter-scanner variability of the HU-density calibration curve was investigated. Results showed that inter-scanner variability was negligible. The median values of individual thresholds th1 (between aerated and intermediate components) were −768, −780, and −798 HU for the three methods, respectively. A significantly lower median value for th2 (between intermediate and consolidated components) was found for the maximum gradient on the data (−34 HU) compared to the other two methods (−114 and −87 HU). The maximum gradient on the data method was applied to quantify the three components in our population—the aerated, intermediate, and consolidation components showed median values of 793 ± 499 cc, 914 ± 291 cc, and 126 ± 111 cc, respectively, while the median value of the first peak was −853 ± 56 HU.

Purpose: To train and validate a predictive model of mortality for hospitalized COVID-19 patients based on lung densitometry. Methods: Two-hundred-fifty-one patients with respiratory symptoms underwent CT few days after hospitalization. "Aerated" (AV), "consolidated" (CV) and "intermediate" (IV) lung sub-volumes were quantified by an operator-independent method based on individual HU maximum gradient recognition. AV, CV, IV, CV/AV, IV/AV, and HU of the first peak position were extracted. Relevant clinical parameters were prospectively collected. The population was composed by training (n = 166) and validation (n = 85) consecutive cohorts, and backward multi-variate logistic regression was applied on the training group to build a CT_model. Similarly, models including only clinical parameters (CLIN_model) and both CT/clinical parameters (COMB_model) were developed. Model's performances were assessed by goodness-of-fit (H&L-test), calibration and discrimination. Model's performances were tested in the validation group. Results: Forty-three patients died (25/18 in training/validation). CT_model included AVmax (i.e. maximum AV between lungs), CV and CV/AE, while CLIN_model included random glycemia, C-reactive protein and biological drugs (protective). Goodness-of-fit and discrimination were similar (H&L:0.70 vs 0.80; AUC:0.80 vs 0.80). COMB_model including AVmax, CV, CV/AE, random glycemia, biological drugs and active cancer, outperformed both models (H & L:0.91; AUC:0.89, 95%CI:0.82-0.93). All models showed good calibration (R-2:0.77-0.97). Despite several patient's characteristics were different between training and validation cohorts, performances in the validation cohort confirmed good calibration (R-2:0-70-0.81) and discrimination for CT_model/COMB_model (AUC:0.72/0.76), while CLIN_model performed worse (AUC:0.64). Conclusions: Few automatically extracted densitometry parameters with clear functional meaning predicted mortality of COVID-19 patients. Combined with clinical features, the resulting predictive model showed higher discrimination/calibration.

One of the most peculiar features of imaging systems is the trade-off between resolution and depth of field. Resolution can be improved by increasing the numerical aperture of the imaging system. However, the range of distances that can be put in sharp focus in a single shot decreases with the square of the numerical aperture. Plenoptic imaging (PI) devices are able to retrieve both spatial and directional information from the scene of interest, usually by placing a microlens array in front of the camera sensor. This feature entails the possibility to refocusing planes of the scene in a much wider range than the natural depth of field of the system, and also to change the point of view on the scene. Though plenoptic imaging is one of the most promising techniques for 3D imaging, its advantages come at the expense of spatial resolution, which can no longer reach the diffraction limit. We experimentally demonstrate that correlations of chaotic light can be exploited to overcome the inverse proportionality between depth of field and resolution, and perform plenoptic imaging at the diffraction limit. We retrieve images by correlating intensity fluctuations at different points of two parts of a sensor, which register spatial and angular information, respectively. Hence, our Correlation Plenoptic Imaging (CPI) protocol does not add any limitation to the native resolution of the imaging system. We show the experimental refocusing, through the CPI procedure, of widely out-of-focus parts of a transmissive test target. Moreover, we determine and test the theoretical limits of CPI in terms of resolution and depth of field, quantifying the improvement with respect to standard imaging and classical PI. We finally comment on future perspectives.

We demonstrate, theoretically and experimentally, the possibility to perform plenoptic imaging by measuring intensity correlations of light. Unlike standard plenoptic procedures, the technique we propose does not sacrifice spatial resolution to achieve directional resolution.

Traditional optical imaging faces an unavoidable trade-off between resolution and depth of field (DOF). To increase resolution, high numerical apertures (NAs) are needed, but the associated large angular uncertainty results in a limited range of depths that can be put in sharp focus. Plenoptic imaging was introduced a few years ago to remedy this trade-off. To this aim, plenoptic imaging reconstructs the path of light rays from the lens to the sensor. However, the improvement offered by standard plenoptic imaging is practical and not fundamental: The increased DOF leads to a proportional reduction of the resolution well above the diffraction limit imposed by the lens NA. In this Letter, we demonstrate that correlation measurements enable pushing plenoptic imaging to its fundamental limits of both resolution and DOF. Namely, we demonstrate maintaining the imaging resolution at the diffraction limit while increasing the depth of field by a factor of 7. Our results represent the theoretical and experimental basis for the effective development of promising applications of plenoptic imaging.

We present the experimental characterization of two distant double-slit masks illuminated by chaotic light, in the absence of first-order imaging and interference. The scheme exploits second-order interference of light propagating through two indistinguishable pairs of disjoint optical paths passing through the masks of interest. The proposed technique leads to a deeper understanding of biphoton interference and coherence, and opens the way to the development of novel schemes for retrieving information on the relative position and the spatial structure of distant objects, which is of interest in remote sensing, biomedical imaging, as well as monitoring of laser ablation, when first-order imaging and interference are not feasible.