Detection of acute rib fractures on CT images with convolutional neural networks: effect of location and type of fracture and reader's experience.

PURPOSE: The evaluation of all ribs on thin-slice CT images is time consuming and it can be difficult to accurately assess the location and type of rib fracture in an emergency. The aim of our study was to develop and validate a convolutional neural network (CNN) algorithm for the detection of acute rib fractures on thoracic CT images and to investigate the effect of the CNN algorithm on radiologists' performance. METHODS: The dataset for development of a CNN consisted of 539 thoracic CT scans with 4906 acute rib fractures. A three-dimensional faster region-based CNN was trained and evaluated by using tenfold cross-validation. For an observer performance study to investigate the effect of CNN outputs on radiologists' performance, 30 thoracic CT scans (28 scans with 90 acute rib fractures and 2 without rib fractures) which were not included in the development dataset were used. Observer performance study involved eight radiologists who evaluated CT images first without and second with CNN outputs. The diagnostic performance was assessed by using figure of merit (FOM) values obtained from the jackknife free-response receiver operating characteristic (JAFROC) analysis. RESULTS: When radiologists used the CNN output for detection of rib fractures, the mean FOM value significantly increased for all readers (0.759 to 0.819, P = 0.0004) and for displaced (0.925 to 0.995, P = 0.0028) and non-displaced fractures (0.678 to 0.732, P = 0.0116). At all rib levels except for the 1st and 12th ribs, the radiologists' true-positive fraction of the detection became significantly increased by using the CNN outputs. CONCLUSION: The CNN specialized for the detection of acute rib fractures on CT images can improve the radiologists' diagnostic performance regardless of the type of fractures and reader's experience. Further studies are needed to clarify the usefulness of the CNN for the detection of acute rib fractures on CT images in actual clinical practice.

Usefulness of Virtual Monochromatic Images and Iodine Maps Derived from Dual-Energy Computed Tomography for Diagnosing Deep Neck Abscesses.

OBJECTIVE: We aimed to determine whether dual-energy computed tomography (CT) is useful for evaluating deep neck abscesses. METHODS: This study included 22 consecutive patients who were clinically suspected of having a deep neck abscess and underwent dual-energy CT. Conventional 120-kVp images, 70- and 40-keV virtual monochromatic images (VMIs), and iodine maps were inspected to calculate the contrast ratio of the abscess rim (AR) to the abscess center (AC) or to the adjacent muscle (M). The diagnostic certainty of abscesses was assessed on these images. RESULTS: Twenty (91%) of 22 patients had a definitive diagnosis. The contrast ratio for AR/AC and AR/M was significantly higher on 40-keV VMIs and iodine maps than on 120-kVp images and 70-keV VMIs (P < 0.05). On both 40-keV VMIs and iodine maps, the diagnostic certainty of abscess improved in 3 (15%) cases compared with 120-kVp images and 70-keV VMIs. CONCLUSIONS: Dual-energy CT-based 40-keV VMIs and iodine maps are useful for evaluating deep neck abscesses and may improve diagnostic certainty.

Evaluation of pituitary structures and lesions with turbo spin-echo diffusion-weighted imaging.

BACKGROUND AND PURPOSE: Turbo spin-echo diffusion-weighted imaging (TSE-DWI) has not been used for evaluating pituitary lesions. We compared the usefulness of TSE-DWI and echo-planar (EP)-DWI for assessing normal pituitary structures and lesions. MATERIALS AND METHODS: Our study included 41 consecutive patients (27 pituitary adenomas, 8 Rathke's cleft cysts, 4 craniopharyngiomas, 1 germinoma, 1 pituitary metastasis) who underwent conventional pre- and post-contrast magnetic resonance imaging (MRI) and TSE- and EP-DWI at 3T. Two observers independently performed qualitative assessment of normal pituitary structures and lesions on sagittal DWI and apparent diffusion coefficient (ADC) maps. One observer recorded ADC values of normal brain structures and pituitary lesions. Kappa (κ) statistics, Wilcoxon signed-rank test, intraclass correlation coefficient, Bland-Altman analysis, Pearson correlation coefficient and independent t-test were used for statistical analysis. RESULTS: Interobserver agreement for qualitative evaluations was good to excellent (κ = 0.65-1.0). On both DWI and ADC maps, visualization of the pituitary gland, of the spatial relationship between the lesion and its normal surroundings, and the whole image quality were significantly better on TSE- than EP sequences (p < .01). In normal brain structures, the ADC value on TSE- and EP-sequences was significantly correlated (r = 0.6979, p < .05). The TSE-ADC value was significantly lower for pituitary adenomas than craniopharyngiomas (p < .05). CONCLUSIONS: For the evaluation of normal pituitary structures and lesions, TSE-DWI is more useful than EP-DWI. The TSE-ADC value may help to differentiate between pituitary adenoma and craniopharyngioma.