Diversity and determinants of the sigmoid septum and its impact on morphology of the outflow tract as revealed using cardiac computed tomography.

BACKGROUND: The sigmoid septum has been generally evaluated subjectively and qualitatively, without detailed examination of its diversity, impact on the morphology of the left ventricular outflow tract (LVOT), and anatomical background. METHODS: We enrolled 100 patients without any background cardiac diseases (67.5 ± 12.8 years old; 43% women) who underwent cardiac computed tomography. Basal septal morphology was evaluated using antero-superior and medial bulging angles (bidirectional angulation of the basal septum relative to the LVOT). The eccentricity index of the LVOT, area narrowing ratio (LVOT/virtual basal ring area), aortic-to-left ventricular axial angle (angulation of the aortic root relative to the left ventricle), and wedged height (non-coronary aortic sinus to inferior epicardium distance) were also quantified. RESULTS: The antero-superior bulging, medial bulging, aortic-to-left ventricular axial angles, LVOT eccentricity index, area narrowing ratio, and wedged height were 76° ± 17°, 166° ± 27°, 127° ± 9°, 1.8 ± 0.5, 1.0 ± 0.2, and 41.2 ± 9.1 mm, respectively. Both bulging angles were correlated with each other and contributed to the narrowing and deformation of the LVOT. Angulated aortic root was not correlated with either bidirectional septal bulge or LVOT narrowing. Clockwise rotation of the aortic root rotation was an independent predictor of prominent antero-superior septal bulge. Deeper aortic wedging was a common independent predictor of bidirectional septal bulge. CONCLUSIONS: The extent of septal bulge varies in normal hearts. Along with deep aortic wedging, the bidirectional bulge of the basal septum deforms and narrows the LVOT without affecting the virtual basal ring morphology.

Deep learning model for predicting gestational age after the first trimester using fetal MRI.

OBJECTIVES: To evaluate a deep learning model for predicting gestational age from fetal brain MRI acquired after the first trimester in comparison to biparietal diameter (BPD). MATERIALS AND METHODS: Our Institutional Review Board approved this retrospective study, and a total of 184 T2-weighted MRI acquisitions from 184 fetuses (mean gestational age: 29.4 weeks) who underwent MRI between January 2014 and June 2019 were included. The reference standard gestational age was based on the last menstruation and ultrasonography measurements in the first trimester. The deep learning model was trained with T2-weighted images from 126 training cases and 29 validation cases. The remaining 29 cases were used as test data, with fetal age estimated by both the model and BPD measurement. The relationship between the estimated gestational age and the reference standard was evaluated with Lin's concordance correlation coefficient (ρc) and a Bland-Altman plot. The ρc was assessed with McBride's definition. RESULTS: The ρc of the model prediction was substantial (ρc = 0.964), but the ρc of the BPD prediction was moderate (ρc = 0.920). Both the model and BPD predictions had greater differences from the reference standard at increasing gestational age. However, the upper limit of the model's prediction (2.45 weeks) was significantly shorter than that of BPD (5.62 weeks). CONCLUSIONS: Deep learning can accurately predict gestational age from fetal brain MR acquired after the first trimester. KEY POINTS: • The prediction of gestational age using ultrasound is accurate in the first trimester but becomes inaccurate as gestational age increases. • Deep learning can accurately predict gestational age from fetal brain MRI acquired in the second and third trimester. • Prediction of gestational age by deep learning may have benefits for prenatal care in pregnancies that are underserved during the first trimester.

Normative Aortic Valvar Measurements in Adults Using Cardiac Computed Tomography　- A Potential Guide to Further Sophisticate Aortic Valve-Sparing Surgery.

BACKGROUND: A thorough understanding of the anatomy of the aortic valve is necessary for aortic valve-sparing surgery. Normal valvar dimensions and their relationships in the living heart, however, have yet to be fully investigated in a 3-dimensional fashion.Methods and Results:In total, 123 consecutive patients (66±12 years, Men 63%) who underwent coronary computed tomographic angiography were enrolled. Mid-diastolic morphology of the aortic roots, including height of the interleaflet triangles, geometric height, free margin length of each leaflet, effective height, and coaptation length were measured using multiplanar reconstruction images. Average height of the interleaflet triangle, geometric height, free margin length, effective height, and the coaptation length were 17.3±1.8, 14.7±1.3, 32.6±3.6, 8.6±1.4, and 3.2±0.8 mm, respectively. The right coronary aortic leaflet displayed the longest free margin length and shortest geometric height. Geometric height, free margin length, and effective height showed positive correlations with aortic root dimensions. Coaptation length, however, remained constant regardless of aortic root dimensions. CONCLUSIONS: Diversities, as well as characteristic relationships among each value involving the aortic root, were identified using living-heart datasets. The aortic leaflets demonstrated compensatory elongation along with aortic root dilatation to maintain constant coaptation length. These measurements will serve as the standard value for revealing the underlying mechanism of aortic regurgitation to plan optimal aortic valve-sparing surgery.

Revisiting the prevalence and diversity of localized thinning of the left ventricular apex.

BACKGROUND: The left ventricular apex commonly has a paper-thin structure. However, available data about its structure are limited to variable samples, methodologies, and results. OBJECTIVE: To investigate the structural anatomy of the left ventricular apex using living heart datasets with the latest computed tomography scanner. METHODS: One hundred thirty-one consecutive patients (median age, 73 years; 58% men) who underwent cardiac computed tomography were retrospectively analyzed. Patients with severe aortic stenosis were analyzed separately. Thickness and diameters of the thinnest part of the left ventricular apex during mid-diastole were measured using orthogonal multiplanar reconstruction images. The area of thinning was estimated using the formula for the ellipse. RESULTS: In 88 patients without severe aortic stenosis, the median thickness of the thinnest area of the left ventricular apex was only 0.9 mm. Among them, 74%, 99%, and 100% of cases displayed a left ventricular apex thinner than 1.0, 3.0, and 5.0 mm, respectively. The median area of the thinnest region was 5.6 mm2 . In 43 patients with severe aortic stenosis, the median thickness of the thinnest area of the left ventricular apex was 1.2 mm. Among them, 51%, 93%, and 100% of cases displayed a left ventricular apex thinner than 1.0, 3.0, and 5.0 mm, respectively. The median area of the thinnest region was 3.9 mm2 . CONCLUSIONS: Localized thinning of the left ventricular apex is unexceptional, regardless of aortic stenosis with concentric left ventricular hypertrophy, thus highlighting the need for a reappreciation of this feature to avoid inadvertent catastrophic complications.

Does Endovascular Abdominal Aortic Repair Change Psoas Muscle Volume?

BACKGROUND: Because endovascular abdominal aortic repair (EVAR) lowers the lumbar arterial blood flow, we hypothesized that the volume of the psoas muscle decreases after surgery. When internal iliac artery (IIA) embolization is performed, the lumbar arterial blood flow further decreases; therefore, we also hypothesized that the decrease in the volume of the psoas muscle becomes more significant. This study was performed to assess the volume change in the psoas muscle after EVAR. METHODS: Fifty-three consecutive patients who underwent EVAR from January 2016 to December 2016 were included. The psoas muscle volume was measured by preoperative and postoperative computed tomography (CT). Postoperative CT scans were performed 6-12 months after EVAR. Axial CT images with a 2-mm slice thickness were used to measure the psoas muscle volume. Data were transferred to a 3-dimensional workstation, and the psoas muscle volume was measured. RESULTS: In the EVAR group, the volume of the psoas muscle decreased by an average of 5.8 mL (4.6%) from 114.8 ± 32.0 mL preoperatively to 109.0 ± 30.3 mL postoperatively (P < 0.01). There was a significant difference in the change in the psoas muscle volume between patients with and without IIA embolization (embolization group: preoperative 118.1 ± 31.0 mL, postoperative 107.5 ± 29.2 mL, mean volume change rate -8.8%; nonembolization group: preoperative 114.0 ± 32.3 mL, postoperative 109.4 ± 30.7 mL, mean volume change rate -3.6%; P < 0.05). CONCLUSIONS: The psoas muscle volume is reduced with EVAR. Moreover, when the IIA is embolized, the psoas muscle volume is further reduced.

Filtered back projection revisited in low-kilovolt computed tomography angiography: sharp filter kernel enhances visualization of the artery of Adamkiewicz.

PURPOSE: The reliability of assessment of the artery of Adamkiewicz before the aortic repair is highly dependent on the display of the continuity of this artery with the aorta, mainly around the vertebral pedicle, by computed tomography angiography (CTA). We hypothesized that the sharp filter kernel can improve visualization of this continuity of the vessel structure because of its edge enhancement and high-spatial resolution. This study was performed to compare the subjective and objective image quality of spinal CTA reconstructed with sharp and smooth filter kernels. METHODS: We retrospectively reviewed 40 consecutive patients who had undergone 80-kV CTA to detect the artery of Adamkiewicz before aortic repair. We measured the CT number and the contrast-to-noise ratio of the anterior spinal artery to the spinal cord. Furthermore, the continuity of the artery of Adamkiewicz was evaluated using a 3-point scale (2 points, absolute; 0 points, undetectable). RESULTS: CTA with the sharp filter kernel showed a significantly higher CT number and contrast-to-noise ratio of the spinal artery than did CTA with the smooth filter kernel (P < .001 for both). Moreover, the sharp filter kernel showed a significantly higher continuity of the artery of Adamkiewicz with the aorta than did the smooth filter kernel (P < .001). CONCLUSIONS: The sharp filter kernel significantly improved the image quality in low-tube-voltage CTA for the assessment of the artery of Adamkiewicz. Thus, CTA with the sharp filter kernel can generate a high-confidence level in the evaluation of the artery of Adamkiewicz.

Focal Myocardial Damage in Cardiac Sarcoidosis Characterized by Strain Analysis on Magnetic Resonance Tagged Imaging in Comparison with Fluorodeoxyglucose Positron Emission Tomography Accumulation and Magnetic Resonance Late Gadolinium Enhancement.

OBJECTIVE: The aims of this study were to characterize focal myocardial damage of cardiac sarcoidosis by strain analysis and to compare it with late gadolinium enhancement (LGE) and fluorodeoxyglucose (FDG) positron emission tomography. METHODS: We reviewed 208 segments from 13 cardiac sarcoidosis patients and measured the circumferential strain (Ecc) and the strain change per second (Ecc rate). The mean Ecc and Ecc rate values were compared between the FDG(+) and FDG(-), and the LGE(+) and LGE(-) segments using Welch's t test. RESULTS: The peak and max Ecc rates were better in the LGE(-) segments than in the LGE(+) segments (-11.8 vs -8.9%, 40.5 vs 29.7%/s, both P < 0.001). The max Ecc rate was higher in the FDG(-) segments than in the FDG(+) segments (39.2 vs 31.7%/s, P < 0.001), but the peak Ecc did not differ between the FDG(+) and FDG(-) segments (-11.2 vs -10.1%, P = 0.17). CONCLUSIONS: Strain analysis could reveal focal myocardial damage in the FDG(+) or the LGE(+) segments.

The association between wedging of the aorta and cardiac structural anatomy as revealed using multidetector-row computed tomography.

The aortic root is wedged within the cardiac base. The precise extent of aortic wedging, however, and its influence on the surrounding cardiac structures, has not been systematically investigated. We analysed 100 consecutive patients, who underwent coronary arterial computed tomographic angiography. We assessed the extent of aortic wedging by measuring the vertical distance between the non-adjacent aortic sinus and the inferior epicardium. A shorter distance indicates deeper aortic wedging. We assessed the tilt angle and diameter of the ascending aorta, the relative heights of the left atrial roof and the oval fossa, the shape of the proximal right coronary artery, the angle of the aorta relative to the left ventricular axis, and the lung volume. The mean extent of wedging was 42.7 ± 9.8 mm. Multivariate analysis revealed that ageing, male gender, increased body mass index, patients without cardiomyopathy, the extent of tilting and dilation of the ascending aorta, and lung volume were all independent predictors for deeper aortic wedging (R2  = 0.7400, P < 0.0001). The extent of wedging was additionally correlated with a relatively high left atrial roof (R2  = 0.1394, P < 0.0001) and oval fossa (R2  = 0.1713, P < 0.0001), the shepherd's crook shape of the proximal right coronary artery (R2  = 0.2376, P < 0.0001), and the narrowness of the angulation of the root relative to the left ventricular axis (R2  = 0.2544, P < 0.0001). In conclusion, ageing, male gender, obesity, background cardiac disease, aortic tilting and dilation, and lung volume are all correlated with the extent of wedging of the aortic root within the cardiac base.

The differences between bisecting and off-center cuts of the aortic root: The three-dimensional anatomy of the aortic root reconstructed from the living heart.

BACKGROUND: It is axiomatic that the diameter of the virtual basal ring of the aortic root, which is elliptical rather than circular, will differ when assessed using between bisecting as opposed to off-center cuts. Such differences, however, which pertain directly to echocardiographic assessments of the so-called valvar annulus, have yet to be systematically explored. METHODS: We retrospectively analyzed 30 patients undergoing coronary computed tomographic angiography, measuring the virtual basal ring diameter using routine multiplanar reconstructions. We made orthogonal bisecting cuts from the nadir of the hinge of the right coronary aortic leaflet to the center of the opposite inter-leaflet fibrous triangle between the noncoronary and left coronary aortic leaflets. We compared these measurements with orthogonal off-center cuts made through the nadirs of the hinges of the adjacent leaflets. RESULTS: The measured diameter of the virtual basal ring was significantly longer when measured using the bisecting cut as opposed to all off-center cuts (mean difference: 1.35±1.34 mm, P<.0001; 0.77±0.95 mm, P=.0001, respectively). The measured diameters of the sinuses of Valsalva, in contrast, were significantly shorter when measured using the bisecting cut (mean difference: -3.24±1.38 mm, P<.0001; -2.86±1.61 mm, P<.0001). CONCLUSIONS: There are significant differences in the diameters of the aortic root, which represent the echocardiographic annulus, when measured using bisecting as opposed to off-center cuts. Account should be taken of these differences when using cross-sectional echocardiographic measurements to assess the dimensions of the aortic root.

Slit-Like Deformation of the Coronary Sinus Orifice due to Compression of the Inferior Pyramidal Space by the Severely Dilated Left Ventricle.

The coronary sinus is located within the inferior pyramidal space, which is the part of the epicardial visceral fibroadipose tissue wedging between the four cardiac chambers from the bottom of the heart. Therefore, this region is susceptible to the morphological changes of the cardiac chambers. We present a case of slit-like deformation of the coronary sinus orifice due to compression of the inferior pyramidal space by the severely dilated left ventricle, which has not been previously described.

Clinical cardiac structural anatomy reconstructed within the cardiac contour using multidetector-row computed tomography: Atrial septum and ventricular septum.

Cardiologists are increasingly becoming involved in procedures associated with the atrial septum and ventricular septum, such as transseptal puncture and selective site pacing. Moreover, detailed knowledge about the architecture of the atrial septum and ventricular septum is now available from studies by radiologists and anatomists. However, from the viewpoint of clinical cardiologists, many questions about the three-dimensional cardiac structural anatomy that relate closely to routine invasive procedures remain unresolved. Although modern multidetector-row computed tomography could provide answers, interventional cardiologists might have not considered the potential of this equipment, as only a few have performed studies with both radiological imaging and cadaveric hearts. Detailed knowledge of the three-dimensional fluoroscopic cardiac structural anatomy could help to reduce the need for contrast medium injection and radiation exposure, and to perform safe interventions. In this article, we present a series of cardiac structural images, including images of the atrial septum and ventricular septum, reconstructed in combination with the cardiac contour using multidetector-row computed tomography. We also discuss the clinical implications of the findings on the basis of accumulated insights of research pioneers. We hope that the present images will serve as a bridge between the fields of cardiology, radiology, and anatomy, and encourage cardiologists to integrate their accumulated insights into the three-dimensional clinical images of the living heart.

Clinical cardiac structural anatomy reconstructed within the cardiac contour using multidetector-row computed tomography: The arrangement and location of the cardiac valves.

It is essential for the interventional cardiologist to have in-depth anatomical information about the three-dimensional arrangement and location of the cardiac valves relative to the various projections of the cardiac contour as revealed fluoroscopically. Multidetector-row computed tomography is useful for providing information about the three-dimensional arrangements of each structure. This article presents cardiac structural images, focusing on the arrangement and location of the cardiac valves, which were reconstructed with the cardiac contour and surrounding structures using multidetector-row computed tomography. We discuss the clinical implications of the findings. We hope these images will serve as a bridge between cardiology, radiology, and anatomy, and will prompt scientists in the field of cardiology to integrate their accumulated insights into three-dimensional clinical images of the living heart.

Clinical cardiac structural anatomy reconstructed within the cardiac contour using multidetector-row computed tomography: Left ventricular outflow tract.

The left ventricular outflow tract (LVOT) is a common site of idiopathic ventricular arrhythmia. Many electrocardiographic characteristics for predicting the origin of arrhythmia have been reported, and their prediction rates are clinically acceptable. Because these approaches are inductive, based on QRS-wave morphology during the arrhythmia and endocardial or epicardial pacing, three-dimensional anatomical accuracy in identifying the exact site of the catheter position is essential. However, fluoroscopic recognition and definition of the anatomy around the LVOT can vary among operators, and three-dimensional anatomical recognition within the cardiac contour is difficult because of the morphological complexity of the LVOT. Detailed knowledge about the three-dimensional fluoroscopic cardiac structural anatomy could help to reduce the need for contrast medium injection and radiation exposure, and to perform safe interventions. In this article, we present a series of structural images of the LVOT reconstructed in combination with the cardiac contour using multidetector-row computed tomography. We also discuss the clinical implications of these findings based on the accumulated insights of research pioneers.

Clinical Structural Anatomy of the Inferior Pyramidal Space Reconstructed Within the Cardiac Contour Using Multidetector-Row Computed Tomography.

Although many studies have described the detailed anatomy of the inferior pyramidal space, it may not be easy for cardiologists who have few chances to study cadaveric hearts to understand the correct morphology of the structure. The inferior pyramidal space is the part of extracardiac fibro-adipose tissue wedging between the 4 cardiac chambers from the diaphragmatic surface of the heart. Many cardiologists have interests in pericardial adipose tissue, but the inferior pyramidal space seems to have been neglected. A number of important structures, including the coronary sinus, atrioventricular node, atrioventricular nodal artery, membranous septum, muscular atrioventricular sandwich (previously called the "muscular atrioventricular septum"), atrial septum, ventricular septum, aortic valvar complex, mitral valvar attachment, and tricuspid valvar attachment are associated with the inferior pyramidal space. We previously revealed its 3-dimensional live anatomy using multidetector-row computed tomography. Moreover, the 3-dimensional understanding of the anatomy in association with the cardiac contour is important from the viewpoints of clinical cardiac electrophysiology. The purpose of this article is to demonstrate extended findings regarding the clinical structural anatomy of the inferior pyramidal space, which was reconstructed in combination with the cardiac contour using multidetector-row computed tomography, and discuss the clinical implications of the findings.

Clinical structural anatomy of the inferior pyramidal space reconstructed from the living heart: Three-dimensional visualization using multidetector-row computed tomography.

The inferior pyramidal space (IPS) comprises the epicardial visceral adipose tissue wedged between the bottoms of the four cardiac chambers from the postero-inferior epicardial surface of the heart. Understanding the complex anatomy around the IPS is important for clinical cardiologists. Although leading anatomists and radiologists have clarified the anatomy of the IPS in detail, few studies have demonstrated this anatomy in three dimensions. The aim of this study was to visualize the three-dimensional anatomy of the IPS reconstructed from the living heart using multidetector-row computed tomography. We also developed an original paper model of the IPS to enhance understanding of its intricate structure.

Optimal angulations for obtaining an en face view of each coronary aortic sinus and the interventricular septum: Correlative anatomy around the left ventricular outflow tract.

An optimal image intensifier angulation used for obtaining an en face view of a target structure is important in electrophysiologic procedures performed around each coronary aortic sinus (CAS). However, few studies have revealed the fluoroscopic anatomy of the target area. This study investigated the optimal angulation for each CAS and the interventricular septum (IVS). The study included 102 consecutive patients who underwent computed tomography coronary angiography. The optimal angle for each CAS was determined by rotating the volume-rendered image around the vertical axis. The angle formed between the anteroposterior axis and IVS was measured using the horizontal section. The frontal direction was defined as zero, positive, or negative if the en face view of the target CAS was obtained in the frontal view, left anterior oblique (LAO) direction, or right anterior oblique (RAO) direction, respectively. The optimal angles for the left, right, and non-CASs were 120.3 ± 10.5°, 4.8 ± 16.3°, and -110.0 ± 13.8°, respectively. The IVS angle was 42.6 ± 8.5°. Accordingly, the optimal image intensifier angulations for the left, right, and non-CASs and the IVS were estimated to be RAO 60°, LAO 5°, LAO 70°, and RAO 50°, respectively. The IVS angle was the most common independent predictor of the optimal angle for each CAS. Differences in the optimal angulations for each CAS and the IVS are demonstrated. The biplane angulation needs to be tailored according to the individual patients and target structures for electrophysiologic procedures.

Association between the rotation and three-dimensional tortuosity of the proximal ascending aorta.

Age-related morphological changes of the aorta, including dilatation and elongation, have been reported. However, rotation has not been fully investigated. We focused on the rotation of the ascending aorta and investigated its relationship with tortuosity. One hundred and two consecutive patients who underwent computed tomography coronary angiography were studied. The angle at which the en face view of the volume-rendered image of the right coronary aortic sinus (RCS) was obtained without foreshortening was defined as the rotation index. It was defined as zero if the RCS was squarely visible in the frontal view, positive if it rotated clockwise toward the left anterior oblique (LAO) direction, and negative if it rotated counter-clockwise toward the right anterior oblique (RAO) direction. The tortuosity was evaluated by measuring the biplane tilt angles formed between the ascending aorta and the horizontal line. The mean rotation index, posterior tilt angle viewed from the RAO direction (αRAO ), and anterior tilt angle viewed from the LAO direction (αLAO ) were 4.8 ± 16.3, 60.7 ± 7.0°, and 63.6 ± 9.0°, respectively. Although no correlation was observed between the rotation index and the αLAO (ß = -0.0761, P = 0.1651), there was a significant negative correlation between the rotation index and αRAO (ß = -0.1810, P < 0.0001). In multivariate regression analysis, the rotation index was an independent predictor of the αRAO (ß = -0.1274, P = 0.0008). Clockwise rotation of the proximal ascending aorta exacerbates the tortuosity by tilting the aorta toward the posterior direction.

Differentiating multiple sclerosis and neuromyelitis optica spectrum disorders through pontine trigeminal nerve lesions: A comparative MRI study.

PURPOSE: Multiple sclerosis (MS) and neuromyelitis optica spectrum disorder (NMOSD) are two major demyelinating diseases affecting the central nervous system (CNS). The objective of this study is to evaluate the prevalence of pontine trigeminal nerve lesions in patients diagnosed with MS and NMOSD using MRI. METHODS: This retrospective study included patients diagnosed with MS or NMOSD between July 2018 and July 2023. MS patients were clinically diagnosed using the 2017 McDonald criteria, while NMOSD patients were those who met the 2015 International Panel for NMO Diagnosis (IPND) criteria and were positive for Aquaporin-4 Antibody (AQP4-Ab). RESULTS: The study included a total of 90 patients, with 45 diagnosed with MS and another 45 with NMOSD. Pontine trigeminal nerve lesions were observed in both MS and NMOSD, but were more prevalent in MS patients (20 % vs. 2 %, p = 0.008). Root entry zone (REZ) lesions were found in 4 of 45 MS patients, accounting for 9 % (95 % CI: 3 %-17 %), and were absent in the NMOSD group; however, there was no significant difference between the two groups (p = 0.12). Of the MS patients with pontine trigeminal nerve lesions, 6 out of 9 (63 %; 95 % CI, 36 %-98 %) exhibited bilateral lesions, which was significantly more prevalent compared to the NMOSD group (13 % vs. 0 %, p = 0.03). CONCLUSIONS: The presence of pontine trigeminal nerve lesions, particularly when bilateral, are significantly more prevalent in MS patients than in those with NMOSD, suggesting their utility as a distinctive marker and potential diagnostic indicator specifically for MS.

Solitary Fibrous Tumor With an Acute Subdural Hematoma: A Case Report and Review of the Literature.

Solitary fibrous tumor (SFT) is a rare interstitial tumor that originates from various soft tissues, and SFTs occurring within the cranium are extremely rare. While intracranial SFTs with cerebral hemorrhage or subarachnoid hemorrhage have been reported, there have been no reports of intracranial SFTs causing subdural hematoma. In this case, we report on an intracranial SFT accompanied by a subdural hematoma. A 29-year-old female was emergently transported due to the sudden onset of persistent headache and vomiting that began the night before. CT and MRI imaging revealed a hemorrhagic tumor under the tentorium and an acute subdural hematoma extending along the tentorium. The excised tumor was diagnosed as an SFT through histopathological examination. After undergoing radiation therapy, no recurrence has been observed. This is the first case report of an SFT accompanied by a subdural hematoma, and it is vital to recognize that SFTs can be associated with subdural hematomas for proper diagnosis and treatment planning.

Prognosis prediction of patients with malignant pleural mesothelioma using conditional variational autoencoder on 3D PET images and clinical data.

BACKGROUND: Deep learning (DL) has been widely used for diagnosis and prognosis prediction of numerous frequently occurring diseases. Generally, DL models require large datasets to perform accurate and reliable prognosis prediction and avoid overlearning. However, prognosis prediction of rare diseases is still limited owing to the small number of cases, resulting in small datasets. PURPOSE: This paper proposes a multimodal DL method to predict the prognosis of patients with malignant pleural mesothelioma (MPM) with a small number of 3D positron emission tomography-computed tomography (PET/CT) images and clinical data. METHODS: A 3D convolutional conditional variational autoencoder (3D-CCVAE), which adds a 3D-convolutional layer and conditional VAE to process 3D images, was used for dimensionality reduction of PET images. We developed a two-step model that performs dimensionality reduction using the 3D-CCVAE, which is resistant to overlearning. In the first step, clinical data were input to condition the model and perform dimensionality reduction of PET images, resulting in more efficient dimension reduction. In the second step, a subset of the dimensionally reduced features and clinical data were combined to predict 1-year survival of patients using the random forest classifier. To demonstrate the usefulness of the 3D-CCVAE, we created a model without the conditional mechanism (3D-CVAE), one without the variational mechanism (3D-CCAE), and one without an autoencoder (without AE), and compared their prediction results. We used PET images and clinical data of 520 patients with histologically proven MPM. The data were randomly split in a 2:1 ratio (train : test) and three-fold cross-validation was performed. The models were trained on the training set and evaluated based on the test set results. The area under the receiver operating characteristic curve (AUC) for all models was calculated using their 1-year survival predictions, and the results were compared. RESULTS: We obtained AUC values of 0.76 (95% confidence interval [CI], 0.72-0.80) for the 3D-CCVAE model, 0.72 (95% CI, 0.68-0.77) for the 3D-CVAE model, 0.70 (95% CI, 0.66-0.75) for the 3D-CCAE model, and 0.69 (95% CI 0.65-0.74) for the without AE model. The 3D-CCVAE model performed better than the other models (3D-CVAE, p = 0.039; 3D-CCAE, p = 0.0032; and without AE, p = 0.0011). CONCLUSIONS: This study demonstrates the usefulness of the 3D-CCVAE in multimodal DL models learned using a small number of datasets. Additionally, it shows that dimensionality reduction via AE can be used to learn a DL model without increasing the overlearning risk. Moreover, the VAE mechanism can overcome the uncertainty of the model parameters that commonly occurs for small datasets, thereby eliminating the risk of overlearning. Additionally, more efficient dimensionality reduction of PET images can be performed by providing clinical data as conditions and ignoring clinical data-related features.