Graphical user interface-based convolutional neural network models for detecting nasopalatine duct cysts using panoramic radiography.

Nasopalatine duct cysts are difficult to detect on panoramic radiographs due to obstructive shadows and are often overlooked. Therefore, sensitive detection using panoramic radiography is clinically important. This study aimed to create a trained model to detect nasopalatine duct cysts from panoramic radiographs in a graphical user interface-based environment. This study was conducted on panoramic radiographs and CT images of 115 patients with nasopalatine duct cysts. As controls, 230 age- and sex-matched patients without cysts were selected from the same database. The 345 pre-processed panoramic radiographs were divided into 216 training data sets, 54 validation data sets, and 75 test data sets. Deep learning was performed for 400 epochs using pretrained-LeNet and pretrained-VGG16 as the convolutional neural networks to classify the cysts. The deep learning system's accuracy, sensitivity, and specificity using LeNet and VGG16 were calculated. LeNet and VGG16 showed an accuracy rate of 85.3% and 88.0%, respectively. A simple deep learning method using a graphical user interface-based Windows machine was able to create a trained model to detect nasopalatine duct cysts from panoramic radiographs, and may be used to prevent such cysts being overlooked during imaging.

Texture analysis using short-tau inversion recovery magnetic resonance images to differentiate squamous cell carcinoma of the gingiva from medication-related osteonecrosis of the jaw.

OBJECTIVES: Despite the difficulty in distinguishing between squamous cell carcinoma (SCC) and medication-related osteonecrosis of the jaw (MRONJ) on the basis of medical imaging examinations, the two conditions have completely different treatment methods and prognoses. Therefore, differentiation of SCC from MRONJ on imaging examinations is very important. This study aimed to distinguish SCC from MRONJ by performing texture analysis using magnetic resonance imaging (MRI) short-tau inversion recovery images. METHODS: This retrospective case-control study included 14 patients with SCC of the lower gingiva and 35 with MRONJ of the mandible who underwent MRI and computed tomography (CT) for suspected SCC or MRONJ. SCC was identified by histopathological examination of tissues excised during surgery. The radiomics features of SCC and MRONJ were analyzed using the open-access software MaZda version 3.3 (Technical University of Lodz, Institute of Electronics, Poland). CT was used to evaluate the presence or absence of qualitative findings (sclerosis, sequestrum, osteolysis, periosteal reaction, and cellulitis) of SCC and MRONJ. RESULTS: Among the 19 texture features selected using MaZda feature-reduction methods, SCC of the gingiva and MRONJ of the mandible revealed differences in two histogram features, one absolute gradient feature, and 16 Gy level co-occurrence matrix features. In particular, the percentile, angular second moment, entropy, and difference entropy exhibited excellent diagnostic performance. CONCLUSION: Non-contrast-enhanced MRI texture analysis revealed differences in texture parameters between mandibular SCC and mandibular MRONJ. MRI texture analysis can be a new noninvasive quantitative method for distinguishing between SCC and MRONJ.

Diagnostic efficacy of apparent diffusion coefficient, texture features, and their combination for differential diagnosis of odontogenic cysts and tumors.

OBJECTIVE: This study assessed the diagnostic efficacy of apparent diffusion coefficient (ADC), texture features, and their combination for the differential diagnosis of odontogenic cysts and tumors with cyst-like features. STUDY DESIGN: In total, 14 dentigerous cysts (DCs), 12 odontogenic keratocysts (OKCs), and 6 unicystic ameloblastomas (UABs) were used as predictor variables in 32 outpatients who underwent magnetic resonance imaging. The outcome variables were ADC, texture features, and their combination for each lesion. Texture features including histogram and gray-level co-occurrence matrix (GLCM) were measured on ADC maps. Ten features were selected by using the Fisher coefficient method. The Kruskal-Wallis test and post hoc Mann-Whitney test with Bonferroni adjustment were used to analyze trivariate statistics. Statistical significance was established at P < .05. Receiver operating characteristic analysis was used to evaluate the diagnostic effect of ADC, texture features, and their combination in distinguishing the lesions from each other. RESULTS: Apparent diffusion coefficient, 1 histogram feature, 9 GLCM features, and their combination demonstrated significant differences between DC, OKC, and UAB (P ≤ .01). Receiver operating characteristic analysis revealed a high area under the curve of .95 to 1.00 for ADC, 10 texture features, and their combination. Sensitivity, specificity, and accuracy ranged from .86 to 1.00. CONCLUSIONS: Apparent diffusion coefficient and texture features, alone or in combination, can be clinically important in facilitating the distinction between these odontogenic lesions.

Risk assessment of external apical root resorption associated with orthodontic treatment using computed tomography texture analysis.

OBJECTIVES: This study aimed to quantitatively assess maxillary central incisor roots using pre-orthodontics computed tomography (CT) texture analysis as part of a radiomics quantitative analysis. METHODS: This retrospective case-control study included 16 patients with external apical root resorption (EARR) and 16 age- and sex-matched patients without EARR, after orthodontic treatment who underwent pre-orthodontics CT for jaw deformities. All patients were treated with a fixed orthodontic appliance before and after surgical orthodontic treatment. EARR was defined as root resorption ≥ 2 mm of the left and right maxillary central incisors on CT images more than 2 years after the start of orthodontic treatment. Texture features of the maxillary central incisor with and without EARR after orthodontic treatment were analyzed using the open-access software, MaZda Ver. 3.3. Ten texture features were selected using the Fisher method in MaZda from 279 original parameters, which were calculated for each of the maxillary central incisors with and without EARR. The results were tested using the Student's t test, Welch's t test, or Mann-Whitney U test. RESULTS: Four gray-level run length matrix features and six gray-level co-occurrence matrix features displayed significant differences between both the groups (p < 0.01). CONCLUSIONS: CT texture analysis was able to quantitatively assess maxillary central incisor roots and distinguish between maxillary central incisor roots with and without EARR. CT texture analysis may be a useful method for predicting EARR after orthodontic treatment.

Texture analysis of low-flow vascular malformations in the oral and maxillofacial region: venous malformation vs. lymphatic malformation.

Purpose: It is challenging for radiologists to distinguish between venous malformations (VMs) and lymphatic malformations (LMs) using magnetic resonance imaging (MRI). Thus, this study aimed to differentiate VMs from LMs using non-contrast-enhanced MRI texture analysis. Material and methods: This retrospective case-control study included 12 LM patients (6 men and 6 women; mean age 43.58, range 7-85 years) and 29 VM patients (7 men and 22 women; mean age 53.10, range 19-76 years) who underwent MRI for suspected vascular malformations. LM and VM patients were identified by histopathological examination of tissues excised during surgery. The texture features of VM and LM were analysed using the open-access software MaZda version 3.3. Seventeen texture features were selected using the Fisher and probability of error and average correlation coefficient methods in MaZda from 279 original parameters calculated for VM and LM. Results: Among 17 selected texture features, the patients with LM and VM revealed significant differences in 1 histogram feature, 8 grey-level co-occurrence matrix (GLCM) features, and 1 grey-level run-length matrix feature. At the cut-off values of the histogram feature [skewness ≤ -0.131], and the GLCM features [S(0, 2) correlation ≥ 0.667, S(0, 3) correlation ≥ 0.451, S(0, 4) correlation ≥ 0.276, S(0, 5) correlation ≥ 0.389, S(1, 1) correlation ≥ 0.739, S(2, 2) correlation ≥ 0.446, S(2, -2) correlation ≥ 0.299, S(3, -3) correlation ≥ 0.091] had area under the curves of 0.724, 0.764, 0.773, 0.747, 0.733, 0.759, 0.730, 0.744 and 0.727, respectively. Conclusions: Non-contrast-enhanced MRI texture analysis allows us to differentiate between LMs and VMs.

Quantitative assessment of the mandibular condyle in patients with diabetes mellitus using diffusion-weighted magnetic resonance imaging.

OBJECTIVE: The purpose of this study was to quantitatively assess the mandibular condyle in patients with diabetes mellitus (DM) using the apparent diffusion coefficient (ADC) on diffusion-weighted magnetic resonance imaging (DWI). STUDY DESIGN: 102 patients with DM and temporomandibular joint (TMJ) pain who underwent magnetic resonance imaging (MRI) examination of the TMJs at our hospital between August 2006 and March 2020 were included in this study. 112 patients with temporomandibular disorders (TMD) who underwent MRI examination at our hospital between April 2019 and March 2020 were included as controls. The MRI findings were compared between the two groups. RESULTS: The mean ADC values of the mandibular condyle in patients with DM were significantly greater than the controls (P < 0.01). Receiver operating characteristic (ROC) curve analysis revealed a cutoff of 0.98 for the ADC values of the mandibular condyle in patients with DM. CONCLUSION: This study found that the ADC on DWI could be used for the quantitative assessment of the mandibular condyle in patients with DM. DWI might serve as a new and noninvasive method to assess the presence of DM.

Risk assessment of lacunar infarct associated with oral conditions: A case control study focused on radiographic bone loss and Eichner classification.

PURPOSE: This study aimed to evaluate whether lacunar infarcts can be predicted from occlusal support and periodontal stage on images. METHODS: Seventy patients with lacunar infarcts and 300 participants without lacunar infarcts who underwent cerebral checkups at our university hospital were retrospectively reviewed. Lacunar infarcts were assessed using magnetic resonance images by a neuroradiologist. The number of teeth, occlusal support, and severity of radiographic bone loss (RBL) were evaluated using computed tomography. Occlusal support was classified according to the Eichner classification. Additionally, patient characteristics were investigated using medical charts and blood test reports. Records of clinical periodontal parameters, such as clinical attachment loss and bleeding on probing, were also evaluated. RESULTS: The severity of RBL and Eichner classification in patients with lacunar infarcts was significantly higher than that in individuals without lacunar infarcts (P<.01). Receiver operating characteristic analysis revealed that the cut-off values for predicting lacunar infarct were ≥ III in the severity of RBL and ≥B1 in the Eichner classification. The corresponding areas under the curve were 0.75 and 0.70, respectively. In multivariate analysis, the factors affecting the severity of RBL (≥ III) (odds ratio [OR], 8.1; 95% confidence interval [CI], 4.1-16.3; P<.001), and Eichner classification (≥B1) (OR, 1.9; 95% CI, 0.86-4.1; P<.05) were significantly associated with the occurrence of lacunar infarcts. CONCLUSION: The severity of RBL and the Eichner classification may be helpful in predicting lacunar infarcts. Therefore, proper periodontal treatment and prosthodontic rehabilitation of missing teeth may prevent lacunar infarcts.

Quantitative assessment of the parotid gland using computed tomography texture analysis to detect parotid sialadenitis.

OBJECTIVE: We aimed to quantitatively assess the parotid gland by using computed tomography (CT) texture analysis to detect parotid sialadenitis (PS). STUDY DESIGN: This retrospective case-control study included 43 patients with PS who underwent CT and magnetic resonance imaging (MRI). Parotid glands with an abnormal signal (STIR: High) on MRI were identified as showing PS. Patients with parotid gland tumors, bilateral PS, marked fatty degeneration, and severe artifacts on CT were excluded. The texture features of parotid glands with PS and the contralateral normal parotid glands were analyzed using the open-access software LIFEx. The regions of interest were manually placed by tracing contours of both parotid glands on CT images. The results were tested with the paired t-test (or Wilcoxon rank-sum test when appropriate). Receiver operating characteristic (ROC) curve analysis was performed to assess the ability of texture features to predict PS. RESULTS: Six gray level run length matrix features, 2 neighborhood gray level difference matrix features, and 5 gray level zone length matrix features displayed significant differences between PS and normal glands (P ≤ .047). ROC curve analysis showed acceptable accuracy in 4 texture features. CONCLUSIONS: CT texture analysis allowed quantitative assessment of parotid glands and may have the potential to detect PS.

Quantitative assessment of mandibular bone marrow using computed tomography texture analysis for detect stage 0 medication-related osteonecrosis of the jaw.

PURPOSE: Medication-related osteonecrosis of the jaw (MRONJ) is a serious complication of treatment with bisphosphonates or antiangiogenic inhibitors. MRONJ has four stages (0-3); however, stage 0 MRONJ is difficult to detect using computed tomography (CT). This study aimed to quantitatively assess the mandibular bone marrow using texture analysis to detect stage 0 MRONJ from CT images. METHODS: This retrospective study included 25 patients with stage 0 MRONJ who had a history of treatment with bisphosphonates and underwent CT and magnetic resonance imaging (MRI). The mandibular bone marrow with abnormal signals (T1-weighted imaging: low, T2-weighted imaging: low or high, short-tau inversion recovery: high) on MRI, and no qualitative characteristic CT and oral findings indicative of osteonecrosis (exposed bone, sequestrum, periosteal reaction, and osteolysis) was identified as 0 MRONJ. Texture features of the bone marrow of the mandible with MRONJ and the contralateral, normal mandibular bone marrow were extracted using an open-access software, namely, LIFEx. The volumes of interest (VOIs) were manually placed on CT images by tracing the bilateral mandibular bone marrow regions, excluding the teeth, mandibular canal, and cortical bone. Thirty-seven texture features were extracted from each VOI. RESULTS: Six gray-level run length matrix features and four gray-level zone length matrix features exhibited significant differences between mandibular bone marrow with and without MRONJ. CONCLUSIONS: CT was able to quantitatively assess texture features of normal mandibular bone marrow and that with MRONJ. Texture analysis may be useful as a new method for detecting stage 0 MRONJ using CT.

Computed tomography texture analysis of mandibular condylar bone marrow in diabetes mellitus patients.

OBJECTIVES: Diabetes mellitus (DM) is associated with a broad range of complications, such as retinopathy, nephropathy, neuropathy, and cardiovascular disease. Therefore, predicting DM from head and neck images is a challenge for clinicians. The purpose of this study was to assess the mandibular condylar bone marrow in DM patients using computed tomography (CT) texture analysis. METHODS: This retrospective study included 16 DM and age and sex matched 16 control patients (11 men, 5 women; mean age, 56.8 ± 14.4 years; range 31-78 years). Patients with Type I DM, prior history of taking bisphosphonates, osteoarthritis of the temporomandibular joint, and CT images with metal artifacts were excluded from this study. Bilateral mandibular condylar bone marrow was manually contoured on axial CT images. The presence or absence of DM is the primary predictor variable. Texture features of the region of interest was the outcome variable, that were analyzed using an open-access software, MaZda Ver.3.3. For each group, 20 features out of 279 parameters were selected with Fisher, probability of error and average correlation coefficient methods in MaZda. Bivariate statistics were computed with the Mann-Whitney U test and the P value was set at .05. RESULTS: One histogram feature, 15 Gy level co-occurrence matrix features, and four gray level run length matrix features showed differences between the DM patients and non-DM patients (P < 0.05). CONCLUSIONS: Several texture features of the condyle demonstrated differences between the DM and non-DM patients. CT texture analysis may potentially detect DM from the condylar bone marrow.

Quantitative assessment of normal submandibular glands and submandibular sialadenitis using CT texture analysis: A retrospective study.

OBJECTIVE: The purpose of this study was to quantitatively assess normal submandibular glands and submandibular sialadenitis (SS) using computed tomography (CT) texture analysis as part of radiomics quantitative analysis. STUDY DESIGN: In total, 31 patients with unilateral SS who underwent head and neck magnetic resonance imaging (MRI) and CT and were retrospectively reviewed. Submandibular glands with abnormal signals (STIR: high, T2-weighted image: high, T1-weighted image: low) on MRI were identified as SS. The radiomics features of the contralateral normal submandibular glands and SS were analyzed using an open-access software, MaZda Version 3.3. Sixteen radiomics features were selected with Fisher and probability of error and average correlation coefficient methods in MaZda from 279 original parameters calculated for each of the normal and SS glands. The results were statistically analyzed with the Wilcoxon rank sum test. RESULTS: One gray-level co-occurrence matrix feature and 9 gray-level run length matrix features displayed significant differences between normal submandibular glands and glands with SS (P < .05). CONCLUSIONS: CT texture analysis was able to quantitatively distinguish between normal and diseased submandibular glands. It therefore may have the potential to detect SS.

Increased Apparent Diffusion Coefficient Values of Masticatory Muscles on Diffusion-Weighted Magnetic Resonance Imaging in Patients With Temporomandibular Joint Disorder and Unilateral Pain.

PURPOSE: Myalgia of the masticatory muscles is difficult to evaluate quantitatively. The purpose of the present study was to quantitatively assess myalgia of the masticatory muscles in patients with temporomandibular disorders (TMDs) using the apparent diffusion coefficient (ADC) on diffusion-weighted magnetic resonance imaging (MRI). PATIENTS AND METHODS: Patients who had undergone MRI with clinically diagnosed TMDs according to the criteria of the American Academy of Orofacial Pain and unilateral temporomandibular joint pain from March 2015 to January 2017 were prospectively enrolled. The MRI techniques used included axial diffusion-weighted imaging (DWI) and short T1 inversion recovery imaging through the neck to the skull base. The regions of interest were drawn to completely include the right and left lateral pterygoid muscles, medial pterygoid muscles, and masseter muscles on a slice demonstrating the largest area of each muscle on the ADC map. We compared each masticatory muscle of the pain side with those of the contralateral side without pain. RESULTS: A total of 106 patients with TMD had met the inclusion criteria (18 males, 88 females; mean age, 48.7 years; range, 16 to 80). The mean ADC values of the masticatory muscles of the pain side were significantly greater than those of the no-pain sides (P < .01), as were those for the lateral pterygoid muscles (1.35 ± 0.79 × 10-3 mm2/second vs 1.13 ± 0.77 × 10-3 mm2/second), medial pterygoid muscles (1.28 ± 0.46 × 10-3 mm2/second vs 1.05 ± 0.69 × 10-3 mm2/second), masseter muscles (1.33 ± 0.78 × 10-3 mm2/second vs 1.09 ± 0.64 × 10-3 mm2/second). CONCLUSIONS: The ADC values of the masticatory muscles on the pain side were significantly greater than those of the contralateral side without pain. Our results suggest that DWI could be used to assess myalgia of the masticatory muscles quantitatively.

Th1 cytokines accentuate but Th2 cytokines attenuate ceramide production in the stratum corneum of human epidermal equivalents: an implication for the disrupted barrier mechanism in atopic dermatitis.

BACKGROUND: Although the mechanism(s) involved in the ceramide deficiency in the stratum corneum (SC) of atopic dermatitis (AD) skin is unknown, Th2 type cytokines have been reported to down-regulate ceramide levels in the epidermis. However, almost all research to date has focused on ceramide levels in the whole epidermis, not just in the SC layers alone, which are predominantly responsible for the skin barrier function. OBJECTIVE: We highlighted the effects of Th1/Th2 cytokines on ceramide levels in the SC. METHODS: We developed a modified system of human epidermal equivalents in which epidermis without a SC is cultured for 1 week to generate complete SC layers after which ceramides are extracted from the separated SC layers and are then quantified as per SC protein. RESULTS: The addition of Th2 cytokines (IL-4/IL-6) at a concentration of 10nM resulted in a marked decrease in SC ceramide levels. The reduced ceramide content in the SC was accompanied by the down-regulated expression of the genes encoding serine-palmitoyl transferase-2, acid sphingomyelinase and ß-glucocerebrosidase in the epidermis. In contrast, the addition of Th1 cytokines (GM-CSF/IFN-γ/TNF-α) at concentrations of 2.5 or 10nM resulted in a slight increase in SC ceramide levels, which were accompanied by no change or an increase in the expression of those genes encoding sphingolipid metabolic enzymes in the epidermis. CONCLUSION: These findings suggest that the Th2 type of inflammation evoked in AD skin is one of the essential factors involved in down-regulating the levels of ceramide in the SC.

A reconstructed human epidermal keratinization culture model to characterize ceramide metabolism in the stratum corneum.

To examine factors that regulate ceramide production during keratinization of the human stratum corneum (SC), we developed a reconstructed human epidermal keratinization model in which a fresh layer of SC is newly formed within 1 week. Addition of the UDP-glucose: ceramide glucosyltransferase inhibitor 1-phenyl-2-decanoylamino-3-morpholino-1-propanol significantly diminished SC ceramide levels (expressed as µg/mg protein) with decreased glucosylceramide levels. Desipramine hydrochloride, an inhibitor of sphingomyelinase, also significantly reduced SC ceramide levels. Similarly, conduritol B epoxide, an inhibitor of ß-glucocerebrosidase, significantly down-regulated SC ceramide levels and significantly increased glucosylceramide levels. These results indicate the reliability of this model to elucidate ceramide synthesis regulating factors. Using this model, we assessed the effects of the inflammatory cytokine interleukin-1α (IL-1α), several bioactive sphingolipids and all-trans retinoic acid (RA) on ceramide levels in the SC. Whereas treatment with IL-1α (at 10 nM) significantly down-regulated ceramide levels, treatment with sphingosylphosphorylcholine (at 50 µM) or sphingosine-1-phosphate (at 10 or 20 µM) distinctly up-regulated ceramide levels. Interestingly, RA (at low as 10 nM) significantly up-regulated ceramide levels without affecting the formation of the SC or levels of keratinization-related proteins in the epidermis. The increased levels of ceramide were accompanied by a significantly increased secretion of granulocyte-macrophage colony-stimulating factor as well as by a significantly down-regulated expression of acid-ceramidase at both the gene and protein levels. Taken together, our results underscore the superiority of this reconstructed human epidermal keratinization model to analyze factors that regulate ceramide synthesis, especially in human SC.

Increase of c-Fos and c-Jun expression in spinal and cranial motoneurons of the degenerating muscle mouse (Scn8a(dmu)).

The degenerating muscle (dmu) mouse harbors a loss-of-function mutation in the Scn8a gene, which encodes the alpha subunit of the voltage-gated sodium channel (VGSC) Na(V)1.6. The distribution of c-Fos and c-Jun was examined in spinal and cranial motoneurons of the dmu mouse. In the cervical spinal cord, trigeminal motor nucleus (Vm), facial nucleus (VII), dorsal motor nucleus of the vagus (X), and hypoglossal nucleus (XII) of wild-type mice, motoneurons expressed c-Fos and c-Jun-immunoreactivity. The immunoreactivity in wild-type mice was mostly weak and localized to the nucleus of these neurons whereas in the spinal cord and brain stem of dmu mice motoneurons showed intense c-Fos and c-Jun-immunoreactivity. The number of c-Fos-immunoreactive motoneurons was dramatically elevated in the cervical spinal cord (wild type, 4.8 +/- 1.0; dmu, 17.3 +/- 1.6), Vm (wild type, 76.2 +/- 21.6; dmu, 216.9 +/- 30.9), VII (wild type, 162.4 +/- 43.3; dmu, 533.3 +/- 41.2), and XII (wild type, 58.2 +/- 43.3; dmu, 150.9 +/- 25.7). The mutation also increased the number of c-Jun-immunoreactive motoneurons in the cervical spinal cord (wild type, 1.6 +/- 0.8; dmu, 12.1 +/- 2.1), Vm (wild type, 41.4 +/- 18.0; dmu, 123.1 +/- 11.7), and X (wild type, 39.1 +/- 10.7; dmu, 92.8 +/- 17.8). The increase of these transcription factors may be associated with the uncoordinated and excessive movement of forelimbs and degeneration of cardiac muscles in dmu mice.