Automated segmentation of trabecular and cortical bone from proton density weighted MRI of the knee

Abstract

Patient-specific implant design and pre- and intra-operative planning is becoming increasingly important in the orthopaedic field. For clinical feasibility of these techniques, fast and accurate segmentation of bone structures from MRI is essential. However, manual segmentation is time intensive and subject to inter- and intra-observer variation. The challenge in developing automatic segmentation algorithms for MRI data mainly exists in the inhomogeneity problem and the low contrast among cortical bone and adjacent tissues. In this paper, we proposed a method for automatic segmentation of knee bone structures for MRI data with a 3D local intensity clustering-based level set and a novel approach to determine the cortical boundary utilizing the normal vector of the trabecular surface. Application to clinical imaging data shows that our method is robust to MRI inhomogeneity. In comparing our method to manual segmentation in 18 femurs and tibiae, we found a dice similarity coefficient (DSC) of 0.9611 ± 0.0052 for the femurs and 0.9591 ± 0.0173 for tibiae. The average surface distance error was 0.4649 ± 0.1430 mm for the femurs and 0.4712 ± 0.2113 mm for the tibiae. The results of the automatic technique thus strongly corresponded to the manual segmentation using less than 3% of the time and with virtually no workload. Graphical abstract ᅟ

Novel Autosegmentation Spatial Similarity Metrics Capture the Time Required to Correct Segmentations Better than Traditional Metrics in a Thoracic Cavity Segmentation Workflow

“…In thoracic cavity segmentations delineated on 329 CT datasets, we evaluated correlations between the time required to review and correct autosegmentations and eight spatial similarity metrics. We find the APL, FNPL, and surface DSC to be better correlates with correction times than traditional metrics, including the ubiquitous 4,6,10,11,16,[22][23][24][25][26]28,29,[32][33][34][35][36][37][38][39][40][41][42][43][44][45][46][47] volumetric DSC. We find that clinical variables that worsen autosegmentation similarity to manually-corrected references do not necessarily prolong the time it takes to correct the autosegmentations.…”

“…In the future, reviewing and vetting autosegmented regions-ofinterest prior to radiomics analyses could become part of routine radiology. 14 Autosegmentations are useful if they obviate the need for a clinician to delineate segmentations de novo, which can be time-consuming 4,[15][16][17][18] and inconsistent [19][20][21][22][23] between observers and within the same observer at different time points. Several studies confirm that clinicians can save time from leveraging autosegmentation templates compared to de novo segmentation, 15,[24][25][26][27][28][29] but in many circumstances the time required for clinicians to review and correct autosegmentations is still substantial.…”

“…Autosegmentations are usually compared with a reference segmentation by spatial similarity metrics that compare 1) volumetric overlap between an autosegmented structure and the same manually segmented structure, 4,6,10,11,16,[22][23][24][25][26]28,29,[32][33][34][35][36][37][38][39][40][41][42][43][44][45][46][47] such as the volumetric Dice similiarity coefficient/index 48 (DSC), or 2) geometric distance between two structures' surfaces, 4,10,22,[24][25][26]28,[32][33][34][35]37,40,42,43,46,47 such as the Hausdorff distance 49 (HD), or 3) structure centricity, 32,39 such as differences between centers-of-mass. Critically, these metrics do not necessarily correlate with time savings in clinical practice.…”

“…Autosegmentation algorithms may soon assist neurologists to localize ischemic cores during a code stroke 1,2 or anticipate Parkinson’s disease onset in an outpatient setting 3 . They may inform 3-D printed implant designs for orthopedists 4,5 or highlight posterior segment lesions 6–8 for ophthalmologists. They may help neurosurgeons spare microvessels, 9 outline catheters for radiation oncologists during MRI-guided brachytherapy, 10 or characterize vocal fold mobility for otorhinolaryngologists 11 .…”

Novel Autosegmentation Spatial Similarity Metrics Capture the Time Required to Correct Segmentations Better Than Traditional Metrics in a Thoracic Cavity Segmentation Workflow

“…Autosegmentation algorithms may soon assist neurologists to localize ischemic cores during a code stroke [1,2] or anticipate Parkinson's disease onset in an outpatient setting [3]. They may inform 3D-printed implant designs for orthopedists [4,5] or highlight posterior segment lesions [6][7][8] for ophthalmologists. They may help neurosurgeons spare microvessels [9], outline catheters for radiation oncologists during MRI-guided brachytherapy [10], or characterize vocal fold mobility for otorhinolaryngologists [11].…”