AI-Based Automated Lipomatous Tumor Segmentation in MR Images: Ensemble Solution to Heterogeneous Data.

Deep learning (DL) has been proposed to automate image segmentation and provide accuracy, consistency, and efficiency. Accurate segmentation of lipomatous tumors (LTs) is critical for correct tumor radiomics analysis and localization. The major challenge of this task is data heterogeneity, including tumor morphological characteristics and multicenter scanning protocols. To mitigate the issue, we aimed to develop a DL-based Super Learner (SL) ensemble framework with different data correction and normalization methods. Pathologically proven LTs on pre-operative T1-weighted/proton-density MR images of 185 patients were manually segmented. The LTs were categorized by tumor locations as distal upper limb (DUL), distal lower limb (DLL), proximal upper limb (PUL), proximal lower limb (PLL), or Trunk (T) and grouped by 80%/9%/11% for training, validation and testing. Six configurations of correction/normalization were applied to data for fivefold-cross-validation trainings, resulting in 30 base learners (BLs). A SL was obtained from the BLs by optimizing SL weights. The performance was evaluated by dice-similarity-coefficient (DSC), sensitivity, specificity, and Hausdorff distance (HD95). For predictions of the BLs, the average DSC, sensitivity, and specificity from the testing data were 0.72 [Formula: see text] 0.16, 0.73 [Formula: see text] 0.168, and 0.99 [Formula: see text] 0.012, respectively, while for SL predictions were 0.80 [Formula: see text] 0.184, 0.78 [Formula: see text] 0.193, and 1.00 [Formula: see text] 0.010. The average HD95 of the BLs were 11.5 (DUL), 23.2 (DLL), 25.9 (PUL), 32.1 (PLL), and 47.9 (T) mm, whereas of SL were 1.7, 8.4, 15.9, 2.2, and 36.6 mm, respectively. The proposed method could improve the segmentation accuracy and mitigate the performance instability and data heterogeneity aiding the differential diagnosis of LTs in real clinical situations.

Total-body PET/CT - First Clinical Experiences and Future Perspectives.

Total-body PET has come a long way from its first conception to today, with both total-body and long axial field of view (> 1m) scanners now being commercially available world-wide. The conspicuous signal collection efficiency gain, coupled with high spatial resolution, allows for higher sensitivity and improved lesion detection, enhancing several clinical applications not readily available on current conventional PET/CT scanners. This technology can provide (a) reduction in acquisition times with preservation of diagnostic quality images, benefitting specific clinical situations (i.e. pediatric patients) and the use of several existing radiotracers that present transient uptake over time and where small differences in acquisition time can greatly impact interpretation of images; (b) reduction in administered activity with minimal impact on image noise, thus reducing effective dose to the patient, improving staff safety, and helping with logistical concerns for short-lived radionuclides or long-lived radionuclides with poor dosimetry profiles that have had limited use on conventional PET scanners until now; (c) delayed scanning, that has shown to increase the detection of even small and previously occult malignant lesions by improved clearance in regions of significant background activity and by reduced visibility of coexisting inflammatory processes; (d) improvement in image quality, as a consequence of higher spatial resolution and sensitivity of total-body scanners, implying better appreciation of small structures and clinical implications with downstream prognostic consequences for patients; (e) simultaneous total-body dynamic imaging, that allows the measurement of full spatiotemporal distribution of radiotracers, kinetic modeling, and creation of multiparametric images, providing physiologic and biologically relevant data of the entire body at the same time. On the other hand, the higher physical and clinical sensitivity of total-body scanners bring along some limitations and challenges. The strong impact on clinical sensitivity potentially increases the number of false positive findings if the radiologist does not recalibrate interpretation considering the new technique. Delayed scanning causes logistical issues and introduces new interpretation questions for radiologists. Data storage capacity, longer processing and reconstruction time issues are other limitations, but they may be overcome in the near future by advancements in reconstruction algorithms and computing hardware.