Dataset Information

Automated selection of mid-height intervertebral disc slice in traverse lumbar spine MRI using a combination of deep learning feature and machine learning classifier.

ABSTRACT: Abnormalities and defects that can cause lumbar spinal stenosis often occur in the Intervertebral Disc (IVD) of the patient's lumbar spine. Their automatic detection and classification require an application of an image analysis algorithm on suitable images, such as mid-sagittal images or traverse mid-height intervertebral disc slices, as inputs. Hence the process of selecting and separating these images from other medical images in the patient's set of scans is necessary. However, the technological progress in making this process automated is still lagging behind other areas in medical image classification research. In this paper, we report the result of our investigation on the suitability and performance of different approaches of machine learning to automatically select the best traverse plane that cuts closest to the half-height of an IVD from a database of lumbar spine MRI images. This study considers images features extracted using eleven different pre-trained Deep Convolution Neural Network (DCNN) models. We investigate the effectiveness of three dimensionality-reduction techniques and three feature-selection techniques on the classification performance. We also investigate the performance of five different Machine Learning (ML) algorithms and three Fully Connected (FC) neural network learning optimizers which are used to train an image classifier with hyperparameter optimization using a wide range of hyperparameter options and values. The different combinations of methods are tested on a publicly available lumbar spine MRI dataset consisting of MRI studies of 515 patients with symptomatic back pain. Our experiment shows that applying the Support Vector Machine algorithm with a short Gaussian kernel on full-length image features extracted using a pre-trained DenseNet201 model is the best approach to use. This approach gives the minimum per-class classification performance of around 0.88 when measured using the precision and recall metrics. The median performance measured using the precision metric ranges from 0.95 to 0.99 whereas that using the recall metric ranges from 0.93 to 1.0. When only considering the L3/L4, L4/L5, and L5/S1 classes, the minimum F1-Scores range between 0.93 to 0.95, whereas the median F1-Scores range between 0.97 to 0.99.

SUBMITTER: Natalia F

PROVIDER: S-EPMC8758114 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:STUDY DESIGN:Prospective case series. OBJECTIVE:Evaluate lumbar paraspinal muscle (PSM) cross-sectional area and intervertebral disc (IVD) height changes induced by a 6-month space mission on the International Space Station. The long-term objective of this project is to promote spine health and prevent spinal injury during space missions and here on Earth. SUMMARY OF BACKGROUND DATA:National Aeronautics and Space Administration (NASA) crewmembers have a 4.3 times higher risk of herniated IVDs, compared with the general and military aviator populations. The highest risk occurs during the first year after a mission. Microgravity exposure during long-duration spaceflights results in approximately 5 cm lengthening of body height, spinal pain, and skeletal deconditioning. How the PSMs and IVDs respond during spaceflight is not well described. METHODS:Six NASA crewmembers were imaged supine with a 3 Tesla magnetic resonance imaging. Imaging was conducted preflight, immediately postflight, and then 33 to 67 days after landing. Functional cross-sectional area (FCSA) measurements of the PSMs were performed at the L3-4 level. FCSA was measured by grayscale thresholding within the posterior lumbar extensors to isolate lean muscle on T2-weighted scans. IVD heights were measured at the anterior, middle, and posterior sections of all lumbar levels. Repeated measures analysis of variance was used to determine significance at P?<?0.05, followed by post-hoc testing. RESULTS:Paraspinal lean muscle mass, as indicated by the FCSA, decreased from 86% of the total PSM cross-sectional area down to 72%, immediately after the mission. Recovery of 68% of the postflight loss occurred during the next 6 weeks, still leaving a significantly lower lean muscle fractional content compared with preflight values. In contrast, lumbar IVD heights were not appreciably different at any time point. CONCLUSION:The data reveal lumbar spine PSM atrophy after long-duration spaceflight. Some FCSA recovery was seen with 46 days postflight in a terrestrial environment, but it remained incomplete compared with preflight levels. LEVEL OF EVIDENCE:4.

Project description:BackgroundDisc height (DH) change is considered one of the most critical factors in assessing intervertebral disc degeneration (IVD). Pfirrmann et al. developed a scoring system for disc degeneration evaluation based on changes in DH in magnetic resonance imaging (MRI). While the relationship between DH measurements and Pfirrmann scores for disc degeneration has been explored, the validity of different DH measuring techniques or their connection with disc degeneration is yet uncertain. The present study investigates intra-rater and inter-rater agreement and reliability of different DH measurement methods on MRI and evaluates the relationship between different DH measurement methods and Pfirrmann scores of IVD degeneration, as well as between different Pfirrmann scores and clinical outcomes.MethodsAdult patients with MRI scans of the lumbar spine were recruited. Eight DH measuring techniques were tested for intra-rater and inter-rater agreement and reliability. Bland and Altman's Limits of Agreement (LOA) was used to evaluate intra-rater and inter-rater agreements. Intra-rater and inter-rater reliability were evaluated using intra-class correlations (ICC) with 95 % confidence intervals (95 % CI). The association between DH and Pfirrmann scores was examined using one-way ANOVA.ResultsExcellent intra-rater reliability was reported for 332 participants on DH (ranging from 0.912 (0.901, 0.923) to 0.973 (0.964, 0.981) and from 0.902 (0.892, 0.915) to 0.975 (0.962, 0.985) by two independent raters). All measuring methods had high intra-rater agreement, except for methods 4 and 5. All methods had good-to-excellent of inter-rater reliability on DH (ICCs ranging from 0.812 (0.795, 0.828) to 0.995 (0.994, 0.995)) except for the posterior disc material length of method 5 (ICC 0.740 (0.718, 0.761)). Methods 1 to 6 for evaluating DH in patients with spondylolisthesis had poor inter-rater reliability. The IVD levels with grades IV and V in Pfirrmann scores had significantly lower DH than the IVD levels with grades I to III in Pfirrmann scores. IVD levels with grades IV and V in Pfirrmann scores had significantly higher VAS and ODI than IVD levels with grades I in Pfirrmann scores.ConclusionA good-to-excellent intra-rater and inter-rater reliability was achieved on most DH measuring methods on MRI following a standardized and structured protocol. However, small anatomical structures and different tissue borders could influence measurements. Additionally, DH can differentiate between grade IV and V Pfirrmann scores, and severe IVD degeneration (IV and V Pfirrmann) is linked to clinical outcomes.

Dataset Information

Automated selection of mid-height intervertebral disc slice in traverse lumbar spine MRI using a combination of deep learning feature and machine learning classifier.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets