Search

Purpose: To elucidate the effect of deep learning–based computer-assisted detection (CAD) on the performance of different-level physicians in detecting intracranial haemorrhage using CT. Methods: A total of 40 head CT datasets (normal, 16; haemorrhagic, 24) were evaluated by 15 physicians (5 board-certificated radiologists, 5 radiology residents, and 5 medical interns). The physicians attended 2 reading sessions without and with CAD. All physicians annotated the haemorrhagic regions with a degree of confidence, and the reading time was recorded in each case. Our CAD system was developed using 433 patients' head CT images (normal, 203; haemorrhagic, 230), and haemorrhage rates were displayed as corresponding probability heat maps using U-Net and a machine learning–based false-positive removal method. Sensitivity, specificity, accuracy, and figure of merit (FOM) were calculated based on the annotations and confidence levels. Results: In patient-based evaluation, the mean accuracy of all physicians significantly increased from 83.7 to 89.7% (p < 0.001) after using CAD. Additionally, accuracies of board-certificated radiologists, radiology residents, and interns were 92.5, 82.5, and 76.0% without CAD and 97.5, 90.5, and 81.0% with CAD, respectively. The mean FOM of all physicians increased from 0.78 to 0.82 (p = 0.004) after using CAD. The reading time was significantly lower when CAD (43 s) was used than when it was not (68 s, p < 0.001) for all physicians. Conclusion: The CAD system developed using deep learning significantly improved the diagnostic performance and reduced the reading time among all physicians in detecting intracranial haemorrhage. [ABSTRACT FROM AUTHOR]

Purpose: To compare the effectiveness of silent susceptibility-weighted angiography (sSWAN), a new imaging technique with lower acoustic noise, with conventional susceptibility-weighted angiography (cSWAN) in the detection of intracranial hemorrhagic lesions. Methods: We measured the acoustic and background noise during sSWAN and cSWAN imaging and calculated the contrast-to-noise ratio (CNR) of the phantom consisting of eight chambers with different concentrations of superparamagnetic iron oxide. In the clinical study, we calculated the CNRs of hemorrhagic lesions in 15 patients and evaluated the images for conspicuity and artifact on each sequence and scored them on a 4-point scale. We also evaluated whether hypointense areas observed on sSWAN or cSWAN increased in size from those on T2*-weighted imaging (T2*-WI). Results: Acoustic noise for sSWAN (57.9 ± 0.32 dB [background noise 51.3 dB]) was significantly less than that for cSWAN (89.0 ± 0.22 dB [background noise 50.9 dB]). The CNRs of phantoms for sSWAN were slightly but not significantly lower than those for cSWAN (P = 0.18). The CNRs of hemorrhagic lesions did not show significant differences between sSWAN and cSWAN (P = 0.17). There were no significant differences between sSWAN and cSWAN with respect to the scores for conspicuity, artifact, and change in size of hypointense areas from T2*-WI. Conclusion: sSWAN is equivalent to cSWAN with respect to the image quality for the detection of hemorrhagic lesions but has lower acoustic noise. [ABSTRACT FROM AUTHOR]