Automated segmentation of magnetic resonance bone marrow signal: a feasibility study.

BACKGROUND: Manual assessment of bone marrow signal is time-consuming and requires meticulous standardisation to secure adequate precision of findings. OBJECTIVE: We examined the feasibility of using deep learning for automated segmentation of bone marrow signal in children and adolescents. MATERIALS AND METHODS: We selected knee images from 95 whole-body MRI examinations of healthy individuals and of children with chronic non-bacterial osteomyelitis, ages 6-18 years, in a longitudinal prospective multi-centre study cohort. Bone marrow signal on T2-weighted Dixon water-only images was divided into three color-coded intensity-levels: 1 = slightly increased; 2 = mildly increased; 3 = moderately to highly increased, up to fluid-like signal. We trained a convolutional neural network on 85 examinations to perform bone marrow segmentation. Four readers manually segmented a test set of 10 examinations and calculated ground truth using simultaneous truth and performance level estimation (STAPLE). We evaluated model and rater performance through Dice similarity coefficient and in consensus. RESULTS: Consensus score of model performance showed acceptable results for all but one examination. Model performance and reader agreement had highest scores for level-1 signal (median Dice 0.68) and lowest scores for level-3 signal (median Dice 0.40), particularly in examinations where this signal was sparse. CONCLUSION: It is feasible to develop a deep-learning-based model for automated segmentation of bone marrow signal in children and adolescents. Our model performed poorest for the highest signal intensity in examinations where this signal was sparse. Further improvement requires training on larger and more balanced datasets and validation against ground truth, which should be established by radiologists from several institutions in consensus.

Quantification of adipose tissues by Dual-Energy X-Ray Absorptiometry and Computed Tomography in colorectal cancer patients.

BACKGROUND & AIMS: Excess adipose tissue may affect colorectal cancer (CRC) patients' disease progression and treatment. In contrast to the commonly used anthropometric measurements, Dual-Energy X-Ray Absorptiometry (DXA) and Computed Tomography (CT) can differentiate adipose tissues. However, these modalities are rarely used in the clinic despite providing high-quality estimates. This study aimed to compare DXA's measurement of abdominal visceral adipose tissue (VAT) and fat mass (FM) against a corresponding volume by CT in a CRC population. Secondly, we aimed to identify the best single lumbar CT slice for abdominal VAT. Lastly, we investigated the associations between anthropometric measurements and VAT estimated by DXA and CT. METHODS: Non-metastatic CRC patients between 50-80 years from the ongoing randomized controlled trial CRC-NORDIET were included in this cross-sectional study. Corresponding abdominal volumes were acquired by Lunar iDXA and from clinically acquired CT examinations. Also, single CT slices at L2-, L3-and L4-level were obtained. Agreement between the methods was investigated using univariate linear regression and Bland-Altman plots. RESULTS: Sixty-six CRC patients were included. Abdominal volumetric VAT and FM measured by DXA explained up to 91% and 96% of the variance in VAT and FM by CT, respectively. Bland-Altman plots demonstrated an overestimation of VAT by DXA compared to CT (mean difference of 76 cm3) concurrent with an underestimation of FM (mean difference of -319 cm3). A higher overestimation of VAT (p = 0.015) and underestimation of FM (p = 0.036) were observed in obese relative to normal weight subjects. VAT in a single slice at L3-level showed the highest explained variance against CT volume (R2 = 0.97), but a combination of three slices (L2, L3, L4) explained a significantly higher variance than L3 alone (R2 = 0.98, p < 0.006). The anthropometric measurements explained between 31-65% of the variance of volumetric VAT measured by DXA and CT. CONCLUSIONS: DXA and the combined use of three CT slices (L2-L4) are valid to predict abdominal volumetric VAT and FM in CRC patients when using volumetric CT as a reference method. Due to the poor performance of anthropometric measurements we recommend exploring the added value of advanced body composition by DXA and CT integrated into CRC care.

Automated Abdominal Segmentation of CT Scans for Body Composition Analysis Using Deep Learning.

Purpose To develop and evaluate a fully automated algorithm for segmenting the abdomen from CT to quantify body composition. Materials and Methods For this retrospective study, a convolutional neural network based on the U-Net architecture was trained to perform abdominal segmentation on a data set of 2430 two-dimensional CT examinations and was tested on 270 CT examinations. It was further tested on a separate data set of 2369 patients with hepatocellular carcinoma (HCC). CT examinations were performed between 1997 and 2015. The mean age of patients was 67 years; for male patients, it was 67 years (range, 29-94 years), and for female patients, it was 66 years (range, 31-97 years). Differences in segmentation performance were assessed by using two-way analysis of variance with Bonferroni correction. Results Compared with reference segmentation, the model for this study achieved Dice scores (mean ± standard deviation) of 0.98 ± 0.03, 0.96 ± 0.02, and 0.97 ± 0.01 in the test set, and 0.94 ± 0.05, 0.92 ± 0.04, and 0.98 ± 0.02 in the HCC data set, for the subcutaneous, muscle, and visceral adipose tissue compartments, respectively. Performance met or exceeded that of expert manual segmentation. Conclusion Model performance met or exceeded the accuracy of expert manual segmentation of CT examinations for both the test data set and the hepatocellular carcinoma data set. The model generalized well to multiple levels of the abdomen and may be capable of fully automated quantification of body composition metrics in three-dimensional CT examinations. © RSNA, 2018 Online supplemental material is available for this article. See also the editorial by Chang in this issue.