Background: Sinus X-ray imaging is still used in the initial evaluation of paranasal sinusitis, which is diagnosed by the opacification or air/fluid level in the sinuses and best seen in the Waters’ view of the paranasal sinus (PNS). The objective of this study was to investigate the feasibility of recognizing the maxillary sinusitis features using PNS X-ray images, as well as to propose the most effective method of determining a reasonable consensus using multiple deep learning models. Methods: A total of 4,860 patients, which included 2,430 normal and maxillary sinusitis subjects each, underwent Waters’ view PNS X-ray scan. The datasets were randomly split into training (70%), validation (15%), and test (15%) subsets. We implemented a majority decision algorithm to determine a reasonable consensus using three multiple convolutional neural network (CNN) models: VGG-16, VGG-19, and ResNet-101. The performance of sinusitis detection was evaluated with quantitative accuracy (ACC) and activation maps. Results: We compared the results of our approaches with ACC and activation maps. ACC [and area under the curve (AUC)] of the internal test dataset was evaluated as 87.4% (0.891), 90.8% (0.891), 93.7% (0.937), and 94.1% (0.948) for VGG-16, VGG-19, ResNet-101, and the majority decision, respectively. ACC (and AUC) of the external test dataset was evaluated as 87.58% (0.877), 87.58% (0.877), 92.12% (0.929), and 94.12% (0.942) for VGG-16, VGG-19, ResNet-101, and the majority decision, respectively. Majority decision algorithms can detect missing and correct lesions using a compensation function of the majority decision. Conclusions: The majority decision algorithm showed high accuracy and significantly more accurate lesion detection compared with those of individual CNN models. The proposed deep learning method with PNS X-ray images can be used as an adjunct to classify maxillary sinusitis.

Purpose: Whiplash injury can cause internal derangement of the temporomandibular joint (TMJ) and lead to temporomandibular disorders (TMDs). Our aim was to evaluate whether the initial clinical findings in TMD patients with whiplash injury are correlated with their magnetic resonance imaging (MRI) characteristics. Materials and methods: This case–control study involved 219 patients (135 women, 84 men; mean age: 37.84 years) who visited our orofacial pain clinic with TMD; TMD was diagnosed using the diagnostic criteria for TMD Axis I. Patients were categorized into three groups based on the presence and type of macrotrauma: in the “wTMD” group, patients had suffered whiplash injury; patients in the “pTMD” group had post-traumatic TMD; the “iTMD” group comprised patients who had presented with TMD symptoms and had sustained no macrotrauma. We investigated the presence of disk displacement, effusion, disk deformity, and condylar degeneration, and changes in the lateral pterygoid muscle (LPM). To evaluate the severity of TMD pain and objectively analyze symptoms, we used a visual analog scale (VAS), palpation index (PI), neck PI, dysfunction index, and craniomandibular index (CMI). Results: The VAS scores, and the severity indexes of the TMD including PI, neck PI, and CMI were highest in the wTMD patients. Atrophy of the LPM was most commonly seen in the wTMD group, as was disk deformity. In wTMD patients only, VAS score was significantly correlated with stress; it was correlated with headache in wTMD and iTMD patients. The clinical symptoms of TMD were not correlated with MRI findings in the wTMD group. However, alterations in the LPM were strongly correlated with disk displacement. Conclusion: If clinicians recognize alterations in the LPM and disk displacement in the TMJ, they will better understand the clinical symptoms and pathophysiology of TMD with whiplash injury. Whiplash injury may lead to TMD via different mechanisms from other macrotraumas.

Purpose: The hyperintense acute reperfusion marker (HARM) is a delayed enhancement of the subarachnoid or subpial space observed on post-contrast fluid-attenuated inversion recovery (FLAIR) images and is associated with permeability changes to the blood–brain barrier in acute stroke. We investigated the relationship between HARM and stroke etiology based on the Trial of ORG 10172 in Acute Stroke Treatment (TOAST) classification. In addition, we evaluated the relationship between HARM and stroke locations with respect to vascular territories and anatomic compartments. Materials and methods: We recruited 264 consecutive patients (109 women; mean age 68.63 years) who were diagnosed with acute ischemic stroke and underwent brain magnetic resonance imaging (MRI) including post-contrast FLAIR and DWI within 7 days of symptom onset from May 2015 to March 2016 for this retrospective study. Post-contrast FLAIR images were obtained 5 min after gadolinium administration. The mean time interval between the onset of stroke symptoms and MRI acquisition in total included patients was 18 h and 7 min (median 12 h and 57 min, range 2–127 h). We analyzed the overall incidence and distribution patterns of HARM in acute ischemic stroke cases and compared the relative incidence and distribution patterns of HARM between the subgroups of stroke etiology based on conventional TOAST classification. We obtained odds ratio (OR) of HARM in different stroke locations based on vascular territories and anatomical compartments. This study was approved by our institutional review board. Results: Among the 264 patients, 67 (25.38%) patients were HARM positive and 197 (74.62%) patients were HARM negative. There was significant difference in HARM incidence among the stroke subgroups (p < 0.001). Small vessel occlusion (SVO) was associated with the HARM-negative group (p < 0.001), while large artery atherosclerosis (LAA) and cardioembolism (CE) were associated with the HARM-positive group (p = 0.001). Also, regional pattern of HARM on the same vascular territory as the acute infarction was dominantly demonstrated regardless of stroke etiology. The OR for HARM from middle cerebral artery (MCA) infarction was 1.868 [95% confidence interval (CI): 1.025–3.401]. The OR for HARM from cortical infarction was 9.475 (95% CI: 4.754–18.883) compared to other anatomic compartments. Conclusion: The presence of the HARM was significantly associated with embolic infarctions including LAA and CE. Conversely, SVO was exclusively associated with the absence of the HARM. Second, MCA and cortical infarction showed a more pronounced HARM compared to infarctions at other vascular territories and anatomic compartments. According to the results in the current study, we speculate that the presence of HARM on post-contrast FLAIR images was associated with specific stroke causes especially in embolic causes.

The purpose of this study was to determine whether a CT interpretation with imaging pattern analysis differentiates Kikuchi disease (KD) from the two more frequently encountered differential lymph nodes diagnoses of tuberculous lymphadenopathy (TL) and reactive hyperplasia (RH). Between January 2012 and July 2015, 20 patients with KD (6 men, 14 women; mean age, 27.80 years), 36 patients with RH (10 men, 26 women; mean age, 33.08 years) and 34 patients with TL (17 men, 17 women; mean age, 39.82 years) were pathologically diagnosed using US-guided fine needle aspiration biopsy, core needle biopsy, or surgical excisional biopsy. We recorded the total number, location, and size of the affected cervical lymph nodes, and two radiologists reviewed the characteristic imaging findings, including the presence of necrosis, cortical enhancement pattern, perinodal infiltration, conglomeration and nodal calcification, to form a consensus. In addition, we compared two attenuation indices on the nonnecrotic portion of the affected lymph nodes, nodal cortical attenuation (NCA) and the ratio of NCA to the adjacent muscle (NCA/M). Conglomeration, enhancement pattern and NCA/M values were independent predictive CT features to distinguish KD from RH. Age and enhancement pattern discriminated KD from TL. Only the mean NCA/M value was a statistically significant CT feature (p = .008) in differentiating KD from both RH+TL. The mean NCA/M of KD (1.67 ± 0.20) was significantly higher than that of RH (1.49 ± 0.20) or TL (1.47 ± 0.21). Our results indicate that in case of nonnecrotic lymphadenopathy, a higher NCA/M index can differentiate KD from RH and TL. In addition, the enhancement pattern according to the degree of necrosis discriminated between KD and TL in the case of necrotic lymphadenopathy.