Deep Learning (DL) technique has shown great potential but still has limited success in online contouring for MR-guided adaptive radiotherapy (MRgART). This study proposed a patient-specific DL auto-segmentation (DLAS) strategy using the patient's previous images and contours to update the model and improve segmentation accuracy and efficiency for MRgART. A prototype model was trained for each patient using the first set of MRI and corresponding contours as inputs. The patient-specific model was updated after each fraction with all the available fractional MRIs/contours, and then used to predict the segmentation for the next fraction. During model training, a variant was fitted under consistency constraints, limiting the differences in the volume, length and centroid between the predictions for the latest MRI within a reasonable range. The model performance was evaluated for both organ-at-risks and tumors auto-segmentation for a total of 6 abdominal/pelvic cases (each with at least 8 sets of MRIs/contours) underwent MRgART through Dice Similarity Coefficient (DSC) and 95% Hausdorff Distance (HD95), and was compared with deformable image registration (DIR) and frozen DL model (no updating after pre-training). The contouring time was also recorded and analyzed. The proposed model achieved superior performance with higher mean DSC (0.90, 95 % CI: 0.88-0.95), as compared to DIR (0.63, 95 %CI: 0.59-0.68) and frozen DL models (0.74, 95 % CI: 0.71-0.79). As for tumors, the proposed method yielded a median DSC of 0.95, 95 % CI: 0.94-0.97, and a median HD95 of 1.63 mm, 95 % CI: 1.22 mm-2.06 mm. The contouring time was reduced significantly (p < 0.05) using the proposed method (73.4 ± 6.5 secs) compared to the manual process (12 ∼ 22 mins). The online ART time was reduced to 1650 ± 274 seconds with the proposed method, as compared to 3251.8 ± 447 seconds using the original workflow. The proposed patient-specific DLAS method can significantly improve the segmentation accuracy and efficiency for longitudinal MRIs, thereby facilitating the routine practice of MRgART.

To perform a dosimetric evaluation of four different simultaneous integrated boost whole brain radiotherapy modalities with hippocampus and inner ear avoidance in the treatment of limited brain metastases. Computed tomography/magnetic resonance imaging data of 10 patients with limited (1–5) brain metastases were used to replan step-and-shoot intensity-modulated radiotherapy (sIMRT), dynamic intensity-modulated radiation therapy (dIMRT), volumetric-modulated arc therapy (VMAT), and helical tomotherapy (Tomo). The prescribed doses of 40–50 Gy in 10 fractions and 30 Gy in 10 fractions were simultaneously delivered to the metastatic lesions and the whole-brain volume, respectively. The hippocampal dose met the RTOG 0933 criteria for hippocampal avoidance (Dmax ≤17 Gy, D100% ≤10 Gy). The inner ear dose was restrained to Dmean ≤15 Gy. Target coverage (TC), homogeneity index (HI), conformity index (CI), maximum dose (Dmax), minimum dose (Dmin) and dose to organs at risk (OARs) were compared. All plans met the indicated dose restrictions. The mean percentage of planning target volume of metastases (PTVmets) coverage ranged from 97.1 to 99.4%. For planning target volume of brain (PTVbrain), Tomo provided the lowest average D2% (37.5 ± 2.8 Gy), the highest average D98% (25.2 ± 2.0 Gy), and the best TC (92.6% ± 2.1%) and CI (0.79 ± 0.06). The two fixed gantry IMRT modalities (step and shot, dynamic) provided similar PTVbrain dose homogeneity (both 0.76). Significant differences across the four approaches were observed for the maximum and minimum doses to the hippocampus and the maximum doses to the eyes, lens and optic nerves. All four radiotherapy modalities produced acceptable treatment plans with good avoidance of the hippocampus and inner ear. Tomo obtained satisfactory PTVbrain coverage and the best homogeneity index. Clinicaltrials.gov, NCT03414944 . Registered 29 January 2018

Orthopedic spine disease is one of the most common diseases in the clinic. The diagnosis of spinal orthopedic injury is an important basis for the treatment of spinal orthopedic diseases. Due to the complexity of the spine structure, doctors usually need to rely on orthopedic CT image data for accurate diagnosis. In some cases, such as poor areas or in emergency situations, it is difficult for doctors to make accurate diagnoses using only 2D x-ray images due to lack of 3D imaging equipment or time crunch. Therefore, an approach based on 2D x-ray images is needed to solve this problem. In this paper, a novel 3D spine reconstruction technique based on 2D orthogonal x-ray images (3DSRNet) is designed. 3DSRNet uses a generative adversarial network architecture and novel modules to make 3D spine reconstruction more accurate and efficient. Spine reconstruction CNN-transformer framework (SRCT) is employed to effectively integrate local bone surface information and long-range relation spinal structure information. Spine reconstruction texture framework (SRTE) is used to extract spine texture features to enhance the effect of pixel-level reconstruction. Experiments show that 3DSRNet achieves excellent 3D spine reconstruction results on multiple metrics including PSNR (45.4666 dB), SSIM (0.8850), CS (0.7662), MAE (23.6696), MSE (9016.1044), and LPIPS (0.0768).

Low-dose computed tomography (LDCT) imaging faces great challenges. Although supervised learning has revealed great potential, it requires sufficient and high-quality references for network training. Therefore, existing deep learning methods have been sparingly applied in clinical practice. To this end, this paper presents a novel Unsharp Structure Guided Filtering (USGF) method, which can reconstruct high-quality CT images directly from low-dose projections without clean references. Specifically, we first employ low-pass filters to estimate the structure priors from the input LDCT images. Then, inspired by classical structure transfer techniques, deep convolutional networks are adopted to implement our imaging method which combines guided filtering and structure transfer. Finally, the structure priors serve as the guidance images to alleviate over-smoothing, as they can transfer specific structural characteristics to the generated images. Furthermore, we incorporate traditional FBP algorithms into self-supervised training to enable the transformation of projection domain data to the image domain. Extensive comparisons and analyses on three datasets demonstrate that the proposed USGF has achieved superior performance in terms of noise suppression and edge preservation, and could have a significant impact on LDCT imaging in the future.