MR Contours Research Articles

Hippocampus segmentation in CT using deep learning: impact of MR versus CT-based training contours.

.Purpose: Hippocampus contouring for radiotherapy planning is performed on MR image data due to poor anatomical visibility on computed tomography (CT) data. Deep learning methods for direct CT hippocampus auto-segmentation exist, but use MR-based training contours. We investigate if these can be replaced by CT-based contours without loss in segmentation performance. This would remove the MR not only from inference but also from training.Approach: The hippocampus was contoured by medical experts on MR and CT data of 45 patients. Convolutional neural networks (CNNs) for hippocampus segmentation on CT were trained on CT-based or propagated MR-based contours. In both cases, their predictions were evaluated against the MR-based contours considered as the ground truth. Performance was measured using several metrics, including Dice score, surface distances, and contour Dice score. Bayesian dropout was used to estimate model uncertainty.Results: CNNs trained on propagated MR contours (median Dice 0.67) significantly outperform those trained on CT contours (0.59) and also experts contouring manually on CT (0.59). Differences between the latter two are not significant. Training on MR contours results in lower model uncertainty than training on CT contours. All contouring methods (manual or CNN) on CT perform significantly worse than a CNN segmenting the hippocampus directly on MR (median Dice 0.76). Additional data augmentation by rigid transformations improves the quantitative results but the difference remains significant.Conclusions: CT-based training contours for CT hippocampus segmentation cannot replace propagated MR-based contours without significant loss in performance. However, if MR-based contours are used, the resulting segmentations outperform experts in contouring the hippocampus on CT.

Feasibility of Deformed MR Contours in Stereotactic Body Radiotherapy of Liver

Feasibility of Deformed MR Contours in Stereotactic Body Radiotherapy of Liver

Head and neck multi-organ auto-segmentation on CT images aided by synthetic MRI.

Because the manual contouring process is labor-intensive and time-consuming, segmentation of organs-at-risk (OARs) is a weak link in radiotherapy treatment planning process. Our goal was to develop a synthetic MR (sMR)-aided dual pyramid network (DPN) for rapid and accurate head and neck multi-organ segmentation in order to expedite the treatment planning process. Forty-five patients' CT, MR, and manual contours pairs were included as our training dataset. Nineteen OARs were target organs to be segmented. The proposed sMR-aided DPN method featured a deep attention strategy to effectively segment multiple organs. The performance of sMR-aided DPN method was evaluated using five metrics, including Dice similarity coefficient (DSC), Hausdorff distance 95% (HD95), mean surface distance (MSD), residual mean square distance (RMSD), and volume difference. Our method was further validated using the 2015 head and neck challenge data. The contours generated by the proposed method closely resemble the ground truth manual contours, as evidenced by encouraging quantitative results in terms of DSC using the 2015 head and neck challenge data. Mean DSC values of 0.91±0.02, 0.73±0.11, 0.96±0.01, 0.78±0.09/0.78±0.11, 0.88±0.04/0.88±0.06 and 0.86±0.08/0.85±0.1 were achieved for brain stem, chiasm, mandible, left/right optic nerve, left/right parotid, and left/right submandibular, respectively. We demonstrated the feasibility of sMR-aided DPN for head and neck multi-organ delineation on CT images. Our method has shown superiority over the other methods on the 2015 head and neck challenge data results. The proposed method could significantly expedite the treatment planning process by rapidly segmenting multiple OARs.

Multi-observer contouring of male pelvic anatomy: Highly variable agreement across conventional and emerging structures of interest.

This study quantified inter-observer contouring variations for multiple male pelvic structures, many of which are of emerging relevance for prostate cancer radiotherapy progression and toxicity response studies. Five prostate cancer patient datasets (CT and T2-weighted MR) were distributed to 13 observers for contouring. CT structures contoured included the clinical target volume (CTV), seminal vesicles, rectum, colon, bowel bag, bladder and peri-rectal space (PRS). MR contours included CTV, trigone, membranous urethra, penile bulb, neurovascular bundle and multiple pelvic floor muscles. Contouring variations were assessed using the intraclass correlation coefficient (ICC), Dice similarity coefficient (DSC), and multiple additional metrics. Clinical target volume (CT and MR), bladder, rectum and PRS contours showed excellent inter-observer agreement (median ICC=0.97; 0.99; 1.00; 0.95; 0.90, DSC=0.83±0.05; 0.88±0.05; 0.93±0.03; 0.81±0.07; 0.80±0.06, respectively). Seminal vesicle contours were more variable (ICC=0.75, DSC=0.73±0.14), while colon and bowel bag contoured volumes were consistent (ICC=0.97; 0.97), but displayed poor overlap (DSC=0.58±0.22; 0.67±0.21). Smaller MR structures showed significant inter-observer variations, with poor overlap for trigone, membranous urethra, penile bulb, and left and right neurovascular bundles (DSC=0.44±0.22; 0.41±0.21; 0.66±0.21; 0.16±0.17; 0.15±0.15). Pelvic floor muscles recorded moderate to strong inter-observer agreement (ICC=0.50-0.97), although large outlier variations were observed. Inter-observer contouring variation was significant for multiple pelvic structures contoured on MR.

Assessment of image quality of a radiotherapy-specific hardware solution for PET/MRI in head and neck cancer patients

Background and purposeFunctional PET/MRI has great potential to improve radiotherapy planning (RTP). However, data integration requires imaging with radiotherapy-specific patient positioning. Here, we investigated the feasibility and image quality of radiotherapy-customized PET/MRI in head-and-neck cancer (HNC) patients using a dedicated hardware setup. Material and methodsTen HNC patients were examined with simultaneous PET/MRI before treatment, with radiotherapy and diagnostic scan setup, respectively. We tested feasibility of radiotherapy-specific patient positioning and compared the image quality between both setups by pairwise image analysis of 18F-FDG-PET, T1/T2-weighted and diffusion-weighted MRI. For image quality assessment, similarity measures including average symmetric surface distance (ASSD) of PET and MR-based tumor contours, MR signal-to-noise ratio (SNR) and mean apparent diffusion coefficient (ADC) value were used. ResultsPET/MRI in radiotherapy position was feasible – all patients were successfully examined. ASSD (median/range) of PET and MR contours was 0.6 (0.4–1.2) and 0.9 (0.5–1.3) mm, respectively. For T2-weighted MRI, a reduced SNR of −26.2% (−39.0–−11.7) was observed with radiotherapy setup. No significant difference in mean ADC was found. ConclusionsSimultaneous PET/MRI in HNC patients using radiotherapy positioning aids is clinically feasible. Though SNR was reduced, the image quality obtained with a radiotherapy setup meets RTP requirements and the data can thus be used for personalized RTP.

Implications of CT imaging for postplan quality assessment in prostate brachytherapy

Postplan quality assurance using CT shows considerable interobserver contour variability. We examined CT postplans of four experienced brachytherapists for comparison with MR-determined prostate volumes. Seventy-five patients had CT and MR scans 1 month post-(125)I prostate brachytherapy. CT scans were contoured by the treating physician and dosimetry calculated. The prostate was contoured independently on MR by one observer with extensive MR experience, the scans were fused and dosimetric parameters compared. The mean prostate volume on CT was 38.3 cc (17.5-78.6 cc), on MR 33.3 cc (16.3-66.1 cc). On average, the volume on CT was 16.1% larger than on MR (range, 8% smaller to 64% larger). Craniocaudal discordance of the CT vs. MR prostate contours ranged from 4 mm cranial to 10 mm caudal to MR base and from 6 mm cranial to 14 mm caudal to MR apex. The CT prostate volume not only included an average of 90% of the MR prostate (range, 75-99%) but also included normal tissue (mean, 8.3 cc; range, 2.9-17.1 cc). The average difference between the calculated D(90) from CT contours vs. MR contours was 10.0 Gy (standard deviation, 8.8; range, -37.6 to +41.6 Gy). On average, only 90% of the MR-defined prostate is included in CT contours, while a volume of normal tissue is erroneously designated as prostate. Lack of awareness of this deficiency in planning and/or operative technique gives a false sense of appreciation of the true conformality, delays implementation of corrective measures, and risks unnecessary side effects.

Effect of geometrical distortion correction in MR on image registration accuracy.

In this article we investigate the effect of geometrical distortion correction in MR images on the accuracy of the registration of X-ray CT and MR head images for both a fiducial marker (extrinsic point) method and a surface-matching technique. We use CT and T2-weighted MR image volumes acquired from seven patients who underwent craniotomies in a stereotactic neurosurgical clinical trial. Each patient had four external markers attached to transcutaneous posts screwed into the outer table of the skull. The MR images are corrected for static field inhomogeneity by using an image rectification technique and corrected for scale distortion (gradient magnitude uncertainty) by using an attached stereotactic frame as an object of known shape and size. We define target registration error (TRE) as the distance between corresponding marker positions after registration and transformation. The accuracy of the fiducial marker method is determined by using each combination of three markers to estimate the transformation and the remaining marker to calculate registration error. Surface-based registration is accomplished by fitting MR contours corresponding to the CSF-dura interface to CT contours derived from the inner surface of the skull. The mean point-based TRE using three noncollinear fiducials improved 34%-from 1.15 to 0.76 mm-after correcting for both static field inhomogeneity and scale distortion. The mean surface-based TRE improved 46%-from 2.20 to 1.19 mm. Correction of geometrical distortion in MR images can significantly improve the accuracy of point-based and surface-based registration of CT and MR head images. Distortion correction can be important in clinical situations such as stereotactic and functional neurosurgery where 1 to 2 mm accuracy is required.