The effect of CT texture-based analysis using machine learning approaches on radiologists' performance in differentiating focal-type autoimmune pancreatitis and pancreatic duct carcinoma.

PURPOSE: To develop a support vector machine (SVM) classifier using CT texture-based analysis in differentiating focal-type autoimmune pancreatitis (AIP) and pancreatic duct carcinoma (PD), and to assess the radiologists' diagnostic performance with or without SVM. MATERIALS AND METHODS: This retrospective study included 50 patients (20 patients with focal-type AIP and 30 patients with PD) who underwent dynamic contrast-enhanced CT. Sixty-two CT texture-based features were extracted from 2D images of the arterial and portal phase CTs. We conducted data compression and feature selections using principal component analysis (PCA) and produced the SVM classifier. Four readers participated in this observer performance study and the statistical significance of differences with and without the SVM was assessed by receiver operating characteristic (ROC) analysis. RESULTS: The SVM performance indicated a high performance in differentiating focal-type AIP and PD (AUC = 0.920). The AUC for all 4 readers increased significantly from 0.827 to 0.911 when using the SVM outputs (p = 0.010). The AUC for inexperienced readers increased significantly from 0.781 to 0.905 when using the SVM outputs (p = 0.310). The AUC for experienced readers increased from 0.875 to 0.912 when using the SVM outputs, however, there was no significant difference (p = 0.018). CONCLUSION: The use of SVM classifier using CT texture-based features improved the diagnostic performance for differentiating focal-type AIP and PD on CT.

Vertebrobasilar Infarction Due to Bow Hunter's Syndrome in a Patient with Rheumatoid Arthritis: A Case Report.

A 60-year-old woman with a 37-year history of rheumatoid arthritis (RA) had a sudden onset of headache. Head MRI showed acute multiple infarctions in the vertebrobasilar region, and MR angiography showed stenosis of the right vertebral artery (VA). 3D-CT angiography of the craniovertebral junction showed atlantoaxial subluxation and stenosis of the right VA just distal to the transverse foramen of C2, which was due to osteophytes and degenerative changes secondary to RA. Digital subtraction angiography clearly demonstrated occlusion of the right VA during rightward head rotation. Based on those findings, rotatory instability at C1-2 was considered as the primary cause of the vertebrobasilar infarctions, and Bow Hunter's syndrome was diagnosed. The patient underwent C1-5 posterior fixation, and brain infarction has not recurred.

The effect of deep convolutional neural networks on radiologists' performance in the detection of hip fractures on digital pelvic radiographs.

PURPOSE: The purpose of our study is to develop deep convolutional neural network (DCNN) for detecting hip fractures using CT and MRI as a gold standard, and to evaluate the diagnostic performance of 7 readers with and without DCNN. METHODS: The study population consisted of 327 patients who underwent pelvic CT or MRI and were diagnosed with proximal femoral fractures. All radiographs were manually checked and annotated by radiologists referring to CT and MRI for selecting ROI. At first, a DCNN with the GoogLeNet model was trained by 302 cases. The remaining 25 cases and 25 control subjects were used for the observer performance study and for the testing of DCNN. Seven readers took part in this study. A continuous rating scale was used to record each observer's confidence level. Subsequently, each observer interpreted with the DCNN outputs and rated them again. The area under the curve (AUC) was used to compare the fracture detection. RESULTS: The average AUC of the 7 readers was 0.832. The AUC of DCNN alone was 0.905. The average AUC of the 7 readers with DCNN outputs was 0.876. The AUC of readers with DCNN output were higher than those without(p < 0.05). The AUC of the 2 experienced readers with DCNN output exceeded the AUC of DCNN alone. CONCLUSION: For detecting the hip fractures on radiographs, DCNN developed using CT and MRI as a gold standard by radiologists improved the diagnostic performance including the experienced readers.

Hippocampal sclerosis without visually detectable hippocampal MRI abnormalities: automated subfield volumetric analysis.

PURPOSE: This study aims to investigate hippocampal subfield volumes in patients with hippocampal sclerosis (HS) without visually detectable MRI abnormalities and to determine the diagnostic accuracy using hippocampal subfield volumes. MATERIALS AND METHODS: We examined 46 patients with unilateral HS who had a histopathological diagnosis, and 54 controls. The patients were divided into two groups; visually detectable HS (n = 26) and undetectable HS (n = 20) on MRI. The volumes of hippocampal subfield using FreeSurfer were compared among the three groups. Diagnostic accuracy was calculated as the AUC of ROC using cutoff values for each individual subfield. RESULTS: Compared with the controls, visually detectable HS showed significantly reduced volumes of all the hippocampal subfields and entire hippocampus, whereas visually undetectable HS showed significant atrophy only in the CA3 and hippocampus-amygdala-transition-area. To diagnose visually undetectable HS, the CA3 volumes had AUC of 0.719, which was higher than AUC of 0.614 based on the entire hippocampal volumes. CONCLUSION: Visually undetectable HS demonstrated volume reductions in the CA3. Further, the CA3 volumes was more useful to diagnose visually undetectable HS compared with the entire hippocampal volumes.

Correction to: The usefulness of full-iterative reconstruction algorithm for the visualization of cystic artery on CT angiography.

The authors wish to replace the Table 1.

Early volume reduction of the hippocampus after whole-brain radiation therapy: an automated brain structure segmentation study.

PURPOSE: To assess atrophy differences among brain regions and time-dependent changes after whole-brain radiation therapy (WBRT). MATERIALS AND METHODS: Twenty patients with lung cancer who underwent both WBRT and chemotherapy (WBRT group) and 18 patients with lung cancer who underwent only chemotherapy (control group) were recruited. Three-dimensional T1WI were analyzed to calculate volume reduction ratio after WBRT in various brain structures. The volume reduction ratio of the hippocampus was compared among following 3 periods: 0-3, 4-7, and 8-11 months after WBRT. RESULTS: The volume reduction ratio of the hippocampus was significantly higher in the WBRT group than in the control group (p < 0.05). In WBRT group, the volume reduction ratio of the hippocampus was significantly higher than that of the cortex and white matter (p < 0.05). There were significant differences in the volume reduction ratio between of 0-3 months and that of 4-7 months (p = 0.02) and between 4-7 months and that of 8-11 months (p = 0.01). CONCLUSION: The hippocampus is more vulnerable to the radiation compared with other brain regions and may become atrophic even in the early stage after WBRT.

The usefulness of full-iterative reconstruction algorithm for the visualization of cystic artery on CT angiography.

PURPOSE: To evaluate the potential of full-iterative reconstruction (IR) for improving image quality of the cystic artery on CT angiography and to assess observer performance. METHODS: Thirty patients who underwent both liver dynamic CT and conventional angiography were included in this retrospective study. All CT data were reconstructed through filtered back projection (FBP), adaptive iterative dose reduction 3D (AIDR3D), and forward-projected, model-based, iterative reconstruction solution (FIRST), respectively. In objective study, we analyzed mean ΔCT numbers (the difference between the HU peak of the vessel and the background) and full-width at tenth-maximum (FWTM) of three parts of the cystic artery by profile curve method comparing the three reconstructions. Subjectively, visualization was evaluated using a four-point scale performed by two blinded observers. ANOVA was used for statistical analysis. RESULTS: In all parts of the cystic artery, the mean ΔCT number of FIRST was shown to be significantly better than that of FBP and AIDR3D (p < 0.05). FWTM in FIRST was the smallest in all of the vessels. The mean visualization score was significantly better with FIRST than with other CT reconstructions (p < 0.05). CONCLUSIONS: The FIRST algorithm led to improved CTA visualization of the cystic artery.