Improved diagnostic performance of plain radiography for cervical ossification of the posterior longitudinal ligament using deep learning.

BACKGROUND: A high false-negative rate has been reported for the diagnosis of ossification of the posterior longitudinal ligament (OPLL) using plain radiography. We investigated whether deep learning (DL) can improve the diagnostic performance of radiologists for cervical OPLL using plain radiographs. MATERIALS AND METHODS: The training set consisted of 915 radiographs from 207 patients diagnosed with OPLL. For the test set, we used 200 lateral cervical radiographs from 100 patients with cervical OPLL and 100 patients without OPLL. An observer performance study was conducted over two reading sessions. In the first session, we compared the diagnostic performance of the DL-model and the six observers. The diagnostic performance was evaluated using the area under the receiver operating characteristic curve (AUC) at the vertebra and patient level. The sensitivity and specificity of the DL model and average observers were calculated in per-patient analysis. Subgroup analysis was performed according to the morphologic classification of OPLL. In the second session, observers evaluated the radiographs by referring to the results of the DL-model. RESULTS: In the vertebra-level analysis, the DL-model showed an AUC of 0.854, which was higher than the average AUC of observers (0.826), but the difference was not significant (p = 0.292). In the patient-level analysis, the performance of the DL-model had an AUC of 0.851, and the average AUC of observers was 0.841 (p = 0.739). The patient-level sensitivity and specificity were 91% and 69% in the DL model, and 83% and 68% for the average observers, respectively. Both the DL-model and observers showed decreases in overall performance in the segmental and circumscribed types. With knowledge of the results of the DL-model, the average AUC of observers increased to 0.893 (p = 0.001) at the vertebra level and 0.911 (p < 0.001) at the patient level. In the subgroup analysis, the improvement was largest in segmental-type (AUC difference 0.087; p = 0.002). CONCLUSIONS: The DL-based OPLL detection model can significantly improve the diagnostic performance of radiologists on cervical radiographs.

Fat quantification of multifidus muscle using T2-weighted Dixon: which measurement methods are best suited for revealing the relationship between fat infiltration and herniated nucleus pulposus.

OBJECTIVE: To reveal the best-suited method for fat quantification of lumbar multifidus to demonstrate its relationship to herniated nucleus pulposus (HNP) using T2-weighted Dixon. MATERIALS AND METHODS: One hundred eight patients who underwent MRI for low back pain were enrolled. Two readers independently analyzed the fat fraction (Ff) using axial two-dimensional (D), coronal 2-D, and coronal 3-D measurement. Pearson's correlation coefficient was calculated between age, body mass index (BMI), and the Ff, and age, sex, BMI, and Ff were compared between 'HNP group' and 'no HNP group'. Multivariate logistic regression analysis was performed to identify factors associated with HNP. RESULTS: Coronal 2-D Ff showed the highest correlation with age (r = 0.536, P < 0.001). Coronal 2-D Ff, and coronal 3-D Ff were significantly higher in those with HNP (coronal 2-D: 18.9 ± 2.9, coronal 3-D: 19.7 ± 2.6, respectively) than those without HNP (coronal 2-D: 17.2 ± 3.2, coronal 3-D: 17.4 ± 3.2, respectively). Ff of all three measurements were significantly higher in those with HNP ≥ 3 levels (axial 2-D: 20.7 ± 3.0, coronal 2-D: 21.1 ± 2.7, coronal 3-D: 21.6 ± 2.5, respectively) than those with HNP <3 levels (axial 2-D: 17.5 ± 4.3, coronal 2-D: 18.5 ± 2.7, coronal 3-D: 19.3 ± 2.5). The BMI was an independent predisposing factor to HNP (P = 0.011). Age and coronal 2-D Ff were significant predictors for multilevel HNP (P = 0.028 and 0.040, respectively). CONCLUSIONS: The Ff of the multifidus muscle on T2-weighted Dixon was associated with age, sex, and HNP. The coronal 2-D measurement was the best suited for fat quantification in multifidus muscle among three measurement methods.

Imaging findings of Stenotrophomonas maltophilia pneumonia: emphasis on CT findings between immunocompromised and immunocompetent patients.

BACKGROUND: Stenotrophomonas maltophilia (S. maltophilia) is a globally emerging, rare, waterborne, aerobic, gram-negative, multiple-drug-resistant organism, most commonly associated with respiratory tract infection in humans. Computed tomography (CT) findings in patients with S. maltophilia pneumonia are rarely reported. PURPOSE: To compare CT findings between immunocompromised and immunocompetent patients, and to determine characteristic imaging findings of S. maltophilia pneumonia. MATERIAL AND METHODS: CT findings of eight immunocompromised and 29 immunocompetent patients with proven S. maltophilia pneumonia were reviewed retrospectively. Different patterns of CT abnormalities between immunocompromised and immunocompetent patients were compared by Fisher's exact test. RESULTS: Patchy ground-glass opacities (GGOs) were the most common CT findings, present in 36 (97.3%) of the 37 patients. Among the patients with patchy GGOs, consolidation was seen in 29 (78.4%) patients, and centrilobular nodules were noted in 15 (40.5%) patients. The transaxial distribution of the parenchymal abnormalities was predominantly randomly distributed in 30 (81.1%) cases. Regarding longitudinal plane involvement, the predominant zonal distributions were the diffuse distribution (n=23, 62.2%) and the lower lung zone (n=14, 37.8%). None of the patients showed upper lung zone predominance. The proportion of patients with parenchymal CT findings or associated findings in the immunocompromised patients was not significantly different from that of the immunocompetent patients. However, lower lung zone predominance on the longitudinal plane was significantly more common in immunocompetent patients than in immunocompromised patients (14/29 vs. 0/8, P=0.015). And diffuse distribution of parenchymal abnormalities on a longitudinal plane was significantly more frequent in immunocompromised patients than in immunocompetent patients (8/8 vs. 15/29, P=0.015). CONCLUSION: The most common CT patterns of S. maltophilia pneumonia in immunocompromised and immunocompetent patients were patchy GGOs and consolidation. However, in immunocompetent patients, parenchymal abnormalities were more predominately distributed in lower lung zone than in immunocompromised patients; and in immunocompromised patients, parenchymal abnormalities were more diffusely distributed than in immunocompetent patients.

Imaging findings of complications after thoracic surgery.

Many surgical procedures are used for the treatment of lung, heart, and mediastinal diseases. The techniques can result in postoperative complications. The nature of these complications differs according to the duration of thoracic surgery and the onset of the complication. The complications occurring within 1 month and more than 1 month generally considered as early and late complications, respectively. Chest radiographs and CT scans obtained in patients who have undergone thoracic surgery show normal changes during the surgical procedure and diverse postsurgical complications. Familiarity with the clinical and radiologic findings of the normal alterations and possible complications after thoracic surgery is crucial in minimizing the increased morbidity and mortality.

Ultrasonography of the pediatric hip and spine.

Ultrasonography (US) is a useful screening method for the diagnosis of developmental dysplasia of the hip (DDH) and congenital spinal anomalies in infants. In addition, US is a useful, noninvasive imaging modality for the diagnosis of transient synovitis in children. The early detection of DDH leads to a better prognosis without surgical intervention. Moreover, spinal US allows the detection of congenital malformations and can also demonstrate normal anatomy and normal variants that may resemble disorders. Therefore, radiologists should be familiar with the sonographic anatomy of the normal infant hip, scanning and measurement techniques for DDH, the US features of transient synovitis in children, spinal US techniques, and the US features of normal anatomical structures, normal variants, and congenital spinal anomalies.