Texture and shape analysis of diffusion-weighted imaging for thyroid nodules classification using machine learning.

PURPOSE: To assess whether the integration between (a) functional imaging features that will be extracted from diffusion-weighted imaging (DWI); and (b) shape and texture imaging features as well as volumetric features that will be extracted from T2-weighted magnetic resonance imaging (MRI) can noninvasively improve the diagnostic accuracy of thyroid nodules classification. PATIENTS AND METHODS: In a retrospective study of 55 patients with pathologically proven thyroid nodules, T2-weighted and diffusion-weighted MRI scans of the thyroid gland were acquired. Spatial maps of the apparent diffusion coefficient (ADC) were reconstructed in all cases. To quantify the nodules' morphology, we used spherical harmonics as a new parametric shape descriptor to describe the complexity of the thyroid nodules in addition to traditional volumetric descriptors (e.g., tumor volume and cuboidal volume). To capture the inhomogeneity of the texture of the thyroid nodules, we used the histogram-based statistics (e.g., kurtosis, entropy, skewness, etc.) of the T2-weighted signal. To achieve the main goal of this paper, a fusion system using an artificial neural network (NN) is proposed to integrate both the functional imaging features (ADC) with the structural morphology and texture features. This framework has been tested on 55 patients (20 patients with malignant nodules and 35 patients with benign nodules), using leave-one-subject-out (LOSO) for training/testing validation tests. RESULTS: The functionality, morphology, and texture imaging features were estimated for 55 patients. The accuracy of the computer-aided diagnosis (CAD) system steadily improved as we integrate the proposed imaging features. The fusion system combining all biomarkers achieved a sensitivity, specificity, positive predictive value, negative predictive value, F1-score, and accuracy of 92.9 % $92.9\%$ (confidence interval [CI]: 78.9 % -- 99.5 % $78.9\%\text{--}99.5\%$ ), 95.8 % $95.8\%$ (CI: 87.4 % -- 99.7 % $87.4\%\text{--}99.7\%$ ), 93 % $93\%$ (CI: 80.7 % -- 99.5 % $80.7\%\text{--}99.5\%$ ), 96 % $96\%$ (CI: 88.8 % -- 99.7 % $88.8\%\text{--}99.7\%$ ), 92.8 % $92.8\%$ (CI: 83.5 % -- 98.5 % $83.5\%\text{--}98.5\%$ ), and 95.5 % $95.5\%$ (CI: 88.8 % -- 99.2 % $88.8\%\text{--}99.2\%$ ), respectively, using the LOSO cross-validation approach. CONCLUSION: The results demonstrated in this paper show the promise that integrating the functional features with morphology as well as texture features by using the current state-of-the-art machine learning approaches will be extremely useful for identifying thyroid nodules as well as diagnosing their malignancy.

Diffusion tensor imaging of vertebral bone marrow in children with Gaucher's disease type I and III: Pre- and post-therapy.

PURPOSE: To assess diffusion tensor imaging (DTI) of the vertebral bone marrow (BM) in children with Gaucher's disease (GD) types I and III before and after therapy. METHODS: Prospective study was conducted upon 25 children with GD type I (n = 17) and III (n = 8) and 13 age and sex-matched controls underwent DTI of vertebral BM. Mean diffusivity (MD) and fractional anisotropy (FA) of vertebral BM was calculated and correlated with genotyping, chitotriosidase, hemoglobin (HB) and, platelet count. RESULTS: There was a statistically significant difference in MD and FA of BM between patients and controls (P = 0.001 and 0.02). The area under the curve (AUC) of MD and FA used to differentiate untreated patients from controls was 0.902 and 0.68 with sensitivity, specificity, and, accuracy 92%, 84.6%, and, 89.5% respectively. There was a significant difference in MD and FA of BM between untreated and treated patients (P = 0.001 and 0.02). AUC of MD and FA used to differentiate untreated from treated patients was 0.93 and 0.649 with sensitivity, specificity, and accuracy of 92%, 80%, and 86% respectively. There was a significant difference in MD and FA (P = 0.03, 0.001 respectively) of BM in GD with homozygous L444P mutation (n = 9) and other mutations (n = 14). Chiotriptase, HB and platelet count of patients was correlated with MD (r = -0.36, 0.42, -0.41) and FA (r = -0.47, -0.37, -0.46) respectively. CONCLUSION: DTI of vertebral BM can help in diagnosis and monitoring patients with GD after therapy and correlated with genotyping, and hematological biomarkers of GD.

Diffusion-weighted magnetic resonance imaging and micro-RNA in the diagnosis of hepatic fibrosis in chronic hepatitis C virus.

BACKGROUND: Diffusion-weighted magnetic resonance imaging has shown promise in the detection and quantification of hepatic fibrosis. In addition, the liver has numerous endogenous micro-RNAs (miRs) that play important roles in the regulation of biological processes such as cell proliferation and hepatic fibrosis. AIM: To assess diffusion-weighted magnetic resonance imaging and miRs in diagnosing and staging hepatic fibrosis in patients with chronic hepatitis C. METHODS: This prospective study included 208 patients and 82 age- and sex-matched controls who underwent diffusion-weighted magnetic resonance imaging of the abdomen, miR profiling, and liver biopsy. Pathological scoring was classified according to the METAVIR scoring system. The apparent diffusion coefficient (ADC) and miR were calculated and correlated with pathological scoring. RESULTS: The ADC value decreased significantly with the progression of fibrosis, from controls (F0) to patients with early fibrosis (F1 and F2) to those with late fibrosis (F3 and F4) (median 1.92, 1.53, and 1.25 × 10-3 mm2/s, respectively) (P = 0.001). The cut-off ADC value used to differentiate patients from controls was 1.83 × 10-3 mm2/s with an area under the curve (AUC) of 0.992. Combining ADC and miR-200b revealed the highest AUC (0.995) for differentiating patients from controls with an accuracy of 96.9%. The cut-off ADC used to differentiate early fibrosis from late fibrosis was 1.54 × 10-3 mm2/s with an AUC of 0.866. The combination of ADC and miR-200b revealed the best AUC (0.925) for differentiating early fibrosis from late fibrosis with an accuracy of 80.2%. The ADC correlated with miR-200b (r = - 0.61, P = 0.001), miR-21 (r = - 0.62, P = 0.001), and miR-29 (r = 0.52, P = 0.001). CONCLUSION: Combining ADC and miRs offers an alternative surrogate non-invasive diagnostic tool for diagnosing and staging hepatic fibrosis in patients with chronic hepatitis C.

Diffusion tensor imaging parameters in differentiation recurrent breast cancer from post-operative changes in patients with breast-conserving surgery.

AIM OF THE WORK: To investigate mean diffusivity (MD) and fractional anisotropy (FA) measured by diffusion tensor imaging (DTI) as complementary tools to differentiate recurrent breast cancer from post-operative changes in patients with breast-conserving surgery (BCS). PATIENTS AND METHODS: Prospective study was conducted upon 30 patients with BCS that underwent DTI and dynamic contrast MR imaging. DTI was performed using an axial two-dimensional spin-echo echo-planar imaging sequence. The MD and FA of the lesions were calculated by 2 observers. A single pixel seed isotropic region of interest was placed in the solid part of the tumor on the axial color FA map guided by an enhanced part of the tumor. The final diagnosis was done by biopsy for all patients. RESULTS: The pathological examination proved to be recurrent breast cancer (n = 13) and post-operative changes (n = 17). Recurrent breast cancer had significantly lower MD (P = 0.001, 0.001) and higher FA (P = 0.003, 0.02) than in post-operative changes for both observers respectively. At ROC curve analysis of MD, the AUC was 0.86 and 0.85 by both observers. The threshed MD was (0.86, 0.85 × 10-3 mm2/s) used for differentiation between entities revealed sensitivity (76.9%, 92.3%), specificity (82.4%, 64.7%) and accuracy (80%, 76.7%) of both observers respectively. At ROC curve analysis of FA, the AUC was 0.82 and 0.75 by both observers. The threshold FA (0.82, 0.75) was used for differentiation between entities revealed sensitivity (92.3%, 76.9%), specificity (70.6%, 70.6%) and accuracy of (80.0%, 73.3%) of both observers respectively. There was a strong positive correlation of MD (r = 0.86) and FA (r = 0.73) of both observers. Combined analysis of FA and MD used for differentiation between entities had AUC (0.90, 0.88) revealed sensitivity (92.3%, 92.3%), specificity (82.4%, 70.6%) and accuracy of (86.7%, 80.0%) for both observers respectively. CONCLUSIONS: Combined analysis of MD and FA of DTI may play an important role as a non-invasive method for differentiation recurrent breast cancer from post-operative changes in patients with BCS.