Development and web deployment of prediction model for pulmonary arterial pressure in chronic thromboembolic pulmonary hypertension using machine learning.

BACKGROUND AND PURPOSE: Mean pulmonary artery pressure (mPAP) is a key index for chronic thromboembolic pulmonary hypertension (CTEPH). Using machine learning, we attempted to construct an accurate prediction model for mPAP in patients with CTEPH. METHODS: A total of 136 patients diagnosed with CTEPH were included, for whom mPAP was measured. The following patient data were used as explanatory variables in the model: basic patient information (age and sex), blood tests (brain natriuretic peptide (BNP)), echocardiography (tricuspid valve pressure gradient (TRPG)), and chest radiography (cardiothoracic ratio (CTR), right second arc ratio, and presence of avascular area). Seven machine learning methods including linear regression were used for the multivariable prediction models. Additionally, prediction models were constructed using the AutoML software. Among the 136 patients, 2/3 and 1/3 were used as training and validation sets, respectively. The average of R squared was obtained from 10 different data splittings of the training and validation sets. RESULTS: The optimal machine learning model was linear regression (averaged R squared, 0.360). The optimal combination of explanatory variables with linear regression was age, BNP level, TRPG level, and CTR (averaged R squared, 0.388). The R squared of the optimal multivariable linear regression model was higher than that of the univariable linear regression model with only TRPG. CONCLUSION: We constructed a more accurate prediction model for mPAP in patients with CTEPH than a model of TRPG only. The prediction performance of our model was improved by selecting the optimal machine learning method and combination of explanatory variables.

Rim Enhancement on Contrast-Enhanced CT as a Predictor of Prognosis in Patients with Pancreatic Ductal Adenocarcinoma.

This study investigated the utility of imaging features, such as rim enhancement on contrast-enhanced CT (CECT), in predicting the prognosis of pancreatic ductal adenocarcinoma (PDAC). This retrospective study included 158 patients (84 men; mean age, 68 years) with pathologically confirmed PDAC. The following imaging features were evaluated on CECT by two radiologists: tumor size, tumor attenuation, and the presence of rim enhancement. Cox proportional hazards analysis was performed to identify the imaging and clinicopathological features for predicting disease-free survival (DFS) and overall survival (OS). Pathological features were compared with the presence of rim enhancement. Among the 158 patients, 106 (67%) underwent curative surgery (surgery group) and 52 (33%) received conservative treatment (non-surgery group). Rim enhancement was observed more frequently in the non-surgery group than in the surgery group (44% vs. 20%; p < 0.001). Rim enhancement showed significant associations with shorter DFS and OS in the surgery group (hazard ratios (HRs), 3.03 and 2.99; p < 0.001 and p = 0.003, respectively), whereas tumor size showed significant associations with shorter OS (HR per 1 mm increase, 1.08; p < 0.001). PDACs with rim enhancement showed significant associations with higher histological tumor grades (p < 0.001). PDAC with rim enhancement on CECT could predict poorer prognosis and more aggressive tumor grades.

Pre-emptive Aortic Side Branch Embolization during Endovascular Aneurysm Repair Using the Excluder Stent-Graft System: A Prospective Multicenter study.

PURPOSE: To evaluate the effectiveness and safety of pre-emptive transcatheter arterial embolization (P-TAE) for aortic side branches (ASBs) to prevent Type 2 endoleaks (EL2) before endovascular aneurysm repair (EVAR) using the Excluder stent-graft system (Excluder). MATERIALS AND METHODS: In this prospective, multicenter study, 80 patients (mean age, 79.1 years [SD ± 6.7]; 85.0% were men; mean aneurysmal sac diameter, 48.4 mm [SD ± 7.4]) meeting the eligibility criteria were prospectively enrolled from 9 hospitals. Before EVAR, P-TAE was performed to embolize the patent ASBs originating from the abdominal aortic aneurysm. Contrast-enhanced computed tomography (CT) was performed at 1 month and 6 months after EVAR. The primary endpoint was EL2 incidence at 6 months, and the secondary endpoints were aneurysmal sac diameter changes at 6 and 12 months, P-TAE outcomes, adverse events related to P-TAE, reintervention, and aneurysm-related mortality. RESULTS: All patients successfully underwent P-TAE without serious. Coil embolization was successful in 81.6% of ASBs. EL2 incidence at 6 months was identified in 18 of 70 (25.7%) patients. Aneurysmal sac diameter shrinkage (≥5 mm) was observed in 30.0% of patients at 6 months and in 40.9% at 12 months. Only 1 patient required reintervention for EL2 within 1 year of EVAR; aneurysm-related deaths were not observed. CONCLUSIONS: P-TAE for ASBs before EVAR using Excluder is a safe and effective strategy. It aids in achieving early aneurysmal sac shrinkage and reduces EL2 reintervention at 1 year after EVAR.

Side branch embolization before endovascular abdominal aortic aneurysm repair to prevent type II endoleak: A prospective multicenter study.

PURPOSE: The purpose of the study was to evaluate the efficacy and safety of pre-emptive transcatheter arterial embolization (P-TAE) of aortic side branches to prevent type II endoleak in patients with abdominal aortic aneurysm after endovascular abdominal aneurysm repair (EVAR). MATERIALS AND METHODS: This multicenter, prospective, single-arm trial enrolled 100 patients with abdominal aortic aneurysm from nine hospitals between 2018 and 2021. There were 85 men and 15 women, with a mean age of 79.6 ± 6.0 (standard deviation) years (range: 65-97 years). P-TAE was attempted for patent aortic side branches, including the inferior mesenteric artery, lumbar arteries, and other branches. The primary endpoint was late type II endoleak incidence at 6 months post-repair. Secondary endpoints included changes in aneurysmal sac diameter at 6- and 12 months, complications, re-intervention, and aneurysm-related mortality. Aneurysm sac changes at 6- and 12 months was compared between the late and no-late type II endoleak groups. RESULTS: Coil embolization was successful in 80.9% (321/397) of patent aortic side branches, including 86.3% of the inferior mesenteric arteries, 80.3% of lumbar arteries, and 55.6% of other branches without severe adverse events. Late type II endoleak incidence at 6 months was 8.9% (8/90; 95% confidence interval: 3.9-16.8%). Aneurysm sac shrinkage > 5 mm was observed in 41.1% (37/90) and 55.3% (47/85) of the patients at 6- and 12-months post-EVAR, respectively. Patients with late type II endoleak had less aneurysm sac shrinkage than those without type II endoleak at 12 months (-0.2 mm vs. -6.0 mm; P = 0.040). No patients required re-intervention for type II endoleak, and no aneurysm-related mortalities occurred. CONCLUSION: P-TAE is safe and effective in preventing type II endoleak, leading to early sac shrinkage at 12 months following EVAR.

A Case of Life-threatening Rupture of Small Renal Angiomyolipoma with an Unidentified Intratumoral Aneurysm during Follow-up.

We report a case of a life-threatening ruptured renal angiomyolipoma (AML) that did not meet the criteria for prophylactic treatment (tumor >4 cm or intratumoral aneurysm >5 mm) during follow-up. A woman in her 70s was followed up for a 2.5-cm AML with a rich vascular component. An intratumoral aneurysm >5 mm was not identified for 2 years. She complained of a sudden abdominal pain with hypotension, and contrast-enhanced computed tomography revealed a retroperitoneal hematoma with contrast media extravasation from an intratumoral aneurysm. Emergency transcatheter arterial embolization was successfully performed using N-butyl cyanoacrylate glue. Rupture can occur in small AMLs or in AMLs not identified with intratumoral aneurysms during follow-up. AMLs with a rich vascular component at the kidney surface are more likely to rupture.

Association between tumor morphology and efficacy of atezolizumab plus bevacizumab for advanced hepatocellular carcinoma.

AIM: The IMbrave150 trial revealed that atezolizumab plus bevacizumab (AtezoBv) showed a higher objective response rate (ORR) in patients with advanced hepatocellular carcinoma (HCC). Although conversion therapy after AtezoBv has been recently reported, markers predictive of its efficacy, particularly radiological imaging markers, have not yet been identified. The present study focused on tumor morphological appearance on radiological imaging and evaluated whether it could be associated with AtezoBv efficacy. METHODS: Ninety-five intrahepatic lesions in 74 patients who were given AtezoBv for advanced HCC were recruited for evaluation. The lesions were divided into two groups, simple nodular (SN group) and non-simple nodular (non-SN group), based on the gross morphology on pretreatment imaging, and retrospectively evaluated for treatment response and other relevant clinical outcomes. RESULTS: Assessing the size of individual tumors after treatment, waterfall plots showed that tumor shrinkage in the non-SN group including 56 lesions was higher than that in the SN group comprising 39 lesions. The ORR was significantly higher in the non-SN group (39.3% vs. 15.4%, p = 0.012). Additionally, the median time to nodular progression was longer in the non-SN group (21.0 months vs. 8.1 months, p = 0.119) compared to the SN group. Six patients with non-SN lesions underwent sequential local therapy. CONCLUSIONS: Atezolizumab plus bevacizumab may show increased therapeutic efficacy in patients with tumors with a higher potential for aggressive oncological behavior, such as non-SN lesions. Treatment strategies focusing on conversion therapy may be crucial in patients with non-SN lesions.

Uterine diverticulum mimicking endometriotic cyst of the ovary.

Uterine diverticulum is a rare congenital malformation caused by abnormal fusion of the Müllerian ducts. The diagnosis of uterine diverticulum is difficult, and it is often misdiagnosed as a Müllerian duct anomaly, degenerated uterine fibroid, or ovarian cyst. We herein report a case of uterine diverticulum mimicking an ovarian endometriotic cyst. A multiparous woman with a history of normal vaginal delivery underwent magnetic resonance imaging for investigation of lower abdominal pain and fever. A 155-mm cystic lesion was observed on the ventral side of the uterus. The content of the cyst showed high signal intensity on T1- and T2-weighted images with precipitates of low signal intensity on the dorsal side, suggesting an endometriotic cyst of the ovary. Surgical and pathological findings revealed that the cyst was pedunculated from the anterior uterine body and composed of 3 layers: CD10-positive endometrium, a smooth muscle layer, and serosa. A uterine diverticulum was definitively diagnosed.

Evaluation and Prediction of Post-Hepatectomy Liver Failure Using Imaging Techniques: Value of Gadoxetic Acid-Enhanced Magnetic Resonance Imaging.

Despite improvements in operative techniques and perioperative care, post-hepatectomy liver failure (PHLF) remains the most serious cause of morbidity and mortality after surgery, and several risk factors have been identified to predict PHLF. Although volumetric assessment using imaging contributes to surgical simulation by estimating the function of future liver remnants in predicting PHLF, liver function is assumed to be homogeneous throughout the liver. The combination of volumetric and functional analyses may be more useful for an accurate evaluation of liver function and prediction of PHLF than only volumetric analysis. Gadoxetic acid is a hepatocyte-specific magnetic resonance (MR) contrast agent that is taken up by hepatocytes via the OATP1 transporter after intravenous administration. Gadoxetic acid-enhanced MR imaging (MRI) offers information regarding both global and regional functions, leading to a more precise evaluation even in cases with heterogeneous liver function. Various indices, including signal intensity-based methods and MR relaxometry, have been proposed for the estimation of liver function and prediction of PHLF using gadoxetic acid-enhanced MRI. Recent developments in MR techniques, including high-resolution hepatobiliary phase images using deep learning image reconstruction and whole-liver T1 map acquisition, have enabled a more detailed and accurate estimation of liver function in gadoxetic acid-enhanced MRI.

Evaluation of the clinical behavior of unclassified renal cell carcinoma and its imaging findings on computed tomography and magnetic resonance imaging based on World Health Organization (WHO) 2022.

OBJECTIVES: To ascertain the clinical behaviors of unclassified renal cell carcinoma (RCC) and its characteristic imaging findings on CT and MRI. METHODS: Subjects in this retrospective study were 10 patients who had received a histological diagnosis of unclassified RCC based on World Health Organization (WHO) 2022 and who had undergone CT and/or MRI prior to surgery. In terms of clinical behaviors, TNM classification, stage, postoperative recurrence, time to recurrence, and postoperative survival were evaluated. In terms of imaging findings, tumor size, growth pattern, CT density, dynamic contrast-enhancement (DCE) pattern, internal appearance, presence of a pseudocapsule, and signal intensity on MRI were evaluated. We compared clinical behaviors and imaging findings, and investigated associations between them. RESULTS: One patient could not be followed-up due to death from other causes. Postoperative recurrence was observed in 4 patients, all of whom had Stage 3 RCC. In the remaining 5 patients without recurrence, all 5 patients showed Stage 2 or below. On imaging, unclassified RCC tended to be large (58.7 mm) and solid (100%), and heterogeneous interiors (80%), cystic degeneration (80%) and high intensity on diffusion-weighted imaging (DWI) (71.4%) were common. Comparing patients with and without recurrence, the following findings tended to differ between recurrence and recurrence-free groups: tumor size (73.4 ± 33.9 mm vs. 50.2 ± 33.9 mm, P = 0.286), growth pattern (invasive: 100% vs. 0%, expansive: 0% vs. 100%, P = 0.008 each), DCE pattern (progressive enhancement pattern, 66.7% vs. 0%, washout pattern, 0% vs. 66.7%, P = 0.135 each) and presence of a pseudocapsule (25% vs. 80%, P = 0.167). CONCLUSION: The clinical behavior of unclassified RCC varies widely. Although imaging findings are also variable, findings of large, heterogeneous tumors with cystic degeneration and high intensity on DWI were common. Several imaging findings such as large size, invasive growth, progressive enhancement pattern and no pseudocapsule may enable prediction of prognosis in unclassified RCC.

Magnetic resonance imaging-based risk factors of hepatocellular carcinoma after direct-acting antiviral therapy: A multicenter observational study.

AIM: To determine risk factors associated with hepatocellular carcinoma (HCC) development following direct-acting antiviral (DAA) therapy. METHODS: We enrolled patients with chronic hepatitis C who underwent direct-acting antiviral therapy and achieved sustained virologic response at 12 weeks between 2012 and 2018. Subsequently, patients were followed up. The primary endpoint was the development of HCC or the date of the last follow up when the absence of HCC was confirmed. Uni- and multivariate Cox proportional hazards models were used to identify factors contributing to HCC development, including gadoxetic acid-enhanced magnetic resonance imaging findings. The cumulative incidence rates of HCC development were calculated using the Kaplan-Meier method, and differences between groups were assessed using the log-rank test. RESULTS: The final study cohort comprised 482 patients (median age 70.5 years; 242 men). The median follow-up period was 36.8 months. Among 482 patients, 96 developed HCC (19.9%). The 1-, 3-, and 5-year cumulative rates of HCC development were 4.9%, 18.6%, and 30.5%, respectively. Multivariate analysis revealed that age, male sex, history of HCC, and hepatobiliary phase hypointense nodules without arterial phase hyperenhancement were independent risk factors significantly associated with HCC development (p < 0.001-0.04). The highest risk group included patients with both a history of HCC and the presence of hepatobiliary phase hypointense nodules without arterial phase hyperenhancement (the 1- and 3-year cumulative HCC development rates were 14.2% and 62.2%, respectively). CONCLUSION: History of HCC and presence of hepatobiliary phase hypointense nodules without arterial phase hyperenhancement were strong risk factors for HCC development following direct-acting antiviral therapy.

Development of Pericardial Fat Count Images Using a Combination of Three Different Deep-Learning Models: Image Translation Model From Chest Radiograph Image to Projection Image of Three-Dimensional Computed Tomography.

RATIONALE AND OBJECTIVES: Pericardial fat (PF)-the thoracic visceral fat surrounding the heart-promotes the development of coronary artery disease by inducing inflammation of the coronary arteries. To evaluate PF, we generated pericardial fat count images (PFCIs) from chest radiographs (CXRs) using a dedicated deep-learning model. MATERIALS AND METHODS: We reviewed data of 269 consecutive patients who underwent coronary computed tomography (CT). We excluded patients with metal implants, pleural effusion, history of thoracic surgery, or malignancy. Thus, the data of 191 patients were used. We generated PFCIs from the projection of three-dimensional CT images, wherein fat accumulation was represented by a high pixel value. Three different deep-learning models, including CycleGAN were combined in the proposed method to generate PFCIs from CXRs. A single CycleGAN-based model was used to generate PFCIs from CXRs for comparison with the proposed method. To evaluate the image quality of the generated PFCIs, structural similarity index measure (SSIM), mean squared error (MSE), and mean absolute error (MAE) of (i) the PFCI generated using the proposed method and (ii) the PFCI generated using the single model were compared. RESULTS: The mean SSIM, MSE, and MAE were 8.56 × 10-1, 1.28 × 10-2, and 3.57 × 10-2, respectively, for the proposed model, and 7.62 × 10-1, 1.98 × 10-2, and 5.04 × 10-2, respectively, for the single CycleGAN-based model. CONCLUSION: PFCIs generated from CXRs with the proposed model showed better performance than those generated with the single model. The evaluation of PF without CT may be possible using the proposed method.

Computer-aided diagnosis of chest X-ray for COVID-19 diagnosis in external validation study by radiologists with and without deep learning system.

To evaluate the diagnostic performance of our deep learning (DL) model of COVID-19 and investigate whether the diagnostic performance of radiologists was improved by referring to our model. Our datasets contained chest X-rays (CXRs) for the following three categories: normal (NORMAL), non-COVID-19 pneumonia (PNEUMONIA), and COVID-19 pneumonia (COVID). We used two public datasets and private dataset collected from eight hospitals for the development and external validation of our DL model (26,393 CXRs). Eight radiologists performed two reading sessions: one session was performed with reference to CXRs only, and the other was performed with reference to both CXRs and the results of the DL model. The evaluation metrics for the reading session were accuracy, sensitivity, specificity, and area under the curve (AUC). The accuracy of our DL model was 0.733, and that of the eight radiologists without DL was 0.696 ± 0.031. There was a significant difference in AUC between the radiologists with and without DL for COVID versus NORMAL or PNEUMONIA (p = 0.0038). Our DL model alone showed better diagnostic performance than that of most radiologists. In addition, our model significantly improved the diagnostic performance of radiologists for COVID versus NORMAL or PNEUMONIA.

Comparison between pystan and numpyro in Bayesian item response theory: evaluation of agreement of estimated latent parameters and sampling performance.

Purpose: The purpose of this study is to compare two libraries dedicated to the Markov chain Monte Carlo method: pystan and numpyro. In the comparison, we mainly focused on the agreement of estimated latent parameters and the performance of sampling using the Markov chain Monte Carlo method in Bayesian item response theory (IRT). Materials and methods: Bayesian 1PL-IRT and 2PL-IRT were implemented with pystan and numpyro. Then, the Bayesian 1PL-IRT and 2PL-IRT were applied to two types of medical data obtained from a published article. The same prior distributions of latent parameters were used in both pystan and numpyro. Estimation results of latent parameters of 1PL-IRT and 2PL-IRT were compared between pystan and numpyro. Additionally, the computational cost of the Markov chain Monte Carlo method was compared between the two libraries. To evaluate the computational cost of IRT models, simulation data were generated from the medical data and numpyro. Results: For all the combinations of IRT types (1PL-IRT or 2PL-IRT) and medical data types, the mean and standard deviation of the estimated latent parameters were in good agreement between pystan and numpyro. In most cases, the sampling time using the Markov chain Monte Carlo method was shorter in numpyro than that in pystan. When the large-sized simulation data were used, numpyro with a graphics processing unit was useful for reducing the sampling time. Conclusion: Numpyro and pystan were useful for applying the Bayesian 1PL-IRT and 2PL-IRT. Our results show that the two libraries yielded similar estimation result and that regarding to sampling time, the fastest libraries differed based on the dataset size.

Comparative study of sub-second temporal resolution 4D-MRI and 4D-CT for target motion assessment in a phantom model.

To develop and investigate the feasibility of sub-second temporal resolution volumetric T1-weighted four-dimensional (4D-) MRI in comparison with 4D-CT for respiratory-correlated motion assessment using an MRI/CT-compatible phantom. Sub-second high temporal resolution (0.5 s) gradient-echo T1-weighted 4D-MRI was developed using a volumetric acquisition scheme with compressed sensing. An MRI/CT-compatible motion phantom (simulated liver tumor) with three sinusoidal movements of amplitudes and two respiratory patterns was introduced and imaged with 4D-MRI and 4D-CT to investigate the geometric accuracy of the target movement. The geometric accuracy, including centroid position, volume, similarity index of dice similarity coefficient (DSC), and Hausdorff distance (HD), was systematically evaluated. Proposed 4D-MRI achieved a similar geometric accuracy compared with 4D-CT regarding the centroid position, volume, and similarity index. The observed position differences of the absolute average centroid were within 0.08 cm in 4D-MRI and 0.03 cm in 4D-CT, less than the 1-pixel resolution for each modality. The observed volume difference in 4D-MRI/4D-CT was within 0.73 cm3 (4.5%)/0.29 cm3 (2.1%) for a large target and 0.06 cm3 (11.3%)/0.04 cm3 (11.6%) for a small target. The observed DSC values for 4D-MRI/4D-CT were at least 0.93/0.95 for the large target and 0.83/0.84 for the small target. The maximum HD values were 0.25 cm/0.31 cm for the large target and 0.21 cm/0.15 cm for the small target. Although 4D-CT potentially exhibit superior numerical accuracy in phantom studies, the proposed high temporal resolution 4D-MRI demonstrates sub-millimetre geometric accuracy comparable to that of 4D-CT. These findings suggest that the 4D-MRI technique is a viable option for characterizing motion and generating phase-dependent internal target volumes within the realm of radiotherapy.

Lenvatinib Mitigates Transarterial Embolization-Induced Polarization of Tumor-Associated Macrophages in a Rat Hepatocellular Carcinoma Model.

PURPOSE: To investigate the effect of transarterial embolization (TAE) on macrophage polarization and the modulatory effect of lenvatinib when used in combination with TAE in a rat hepatocellular carcinoma model. MATERIALS AND METHODS: A N1S1-bearing orthotopic rat model was subjected to TAE and administered 5 mg/kg of lenvatinib. CD8+, CD68+, and CD206+ cells were examined in 4 groups: sham (n = 5), lenvatinib (n = 5), TAE (n = 5), and combination of TAE and lenvatinib (n = 5). Transcriptome analysis was performed to assess gene expression related to macrophage polarization in the sham, TAE, and combination groups. An in vitro coculture experiment with bone marrow-derived macrophages was performed to identify lenvatinib target in macrophage polarization. RESULTS: There were no significant differences in the number of CD8+ and CD68+ cells among the 4 groups. Tumor-associated macrophage positivity for CD206 was significantly higher in the TAE group (58.1 ± 20.9) than in the sham (11.2 ± 14.3; P < .001) and combination (27.1 ± 19.7; P = .003) groups. In the transcriptome analysis, compared with the genes in the sham group, 5 macrophage polarization-related genes, including St6gal1, were upregulated by more than 1.5 fold in the TAE group and downregulated by more than 1.5 fold in the combination group. The coculture experiment showed that lenvatinib did not affect macrophages but affected N1S1 cells, leading to macrophage polarization. CONCLUSIONS: TAE-induced M2 macrophage polarization. Lenvatinib administration with TAE could reprogram macrophage polarization, improving tumor immune microenvironment.

Report of the 23rd nationwide follow-up survey of primary liver cancer in Japan (2014-2015).

For the 23rd Nationwide Follow-up Survey of Primary Liver Cancer in Japan, data from 20 889 newly registered patients and 42 274 previously registered follow-up patients were compiled from 516 institutions over a 2-year period from January 1, 2014 to December 31, 2015. Basic statistics compiled for patients newly registered in the 23rd survey were cause of death, past medical history, clinical diagnosis, imaging diagnosis, treatment-related factors, pathological diagnosis, recurrence status, and autopsy findings. Compared with the previous 22nd survey, the population of patients with hepatocellular carcinoma (HCC) was older at the time of clinical diagnosis, had more female patients, had more patients with non-B non-C HCC, had smaller tumor diameter, and was more frequently treated with hepatectomy. Cumulative survival rates were calculated for HCC, intrahepatic cholangiocarcinoma, and combined hepatocellular cholangiocarcinoma (combined HCC and intrahepatic cholangiocarcinoma) by treatment type and background characteristics for patients newly registered between 2004 and 2015 whose final outcome was survival or death. The median overall survival and cumulative survival rates for HCC were calculated by dividing patients by combinations of background factors (number of tumors, tumor diameter, Child-Pugh grade, or albumin-bilirubin grade) and by treatment type (hepatectomy, radiofrequency ablation therapy, transcatheter arterial chemoembolization, hepatic arterial infusion chemotherapy, and systemic therapy). The same values were also calculated according to registration date by dividing patients newly registered between 1978 and 2015 into five time period groups. The data obtained from this nationwide follow-up survey are expected to contribute to advancing clinical research and treatment of primary liver cancer in the world.

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.

Plug-assisted Retrograde Transvenous Obliteration-II for Gastric Varices: A Case Report.

We present a case of gastric varices successfully treated with modified plug-assisted retrograde transvenous obliteration. A 45-year-old male patient had isolated fundal gastric varices caused by alcoholic cirrhosis. Contrast-enhanced computed tomography showed that the gastric varices were drained mainly via the gastro-renal shunt. The gastric varices were treated via plug-assisted retrograde transvenous obliteration using an IMPEDE vascular plug with a modified coil-assisted retrograde transvenous obliteration-II procedure. There were no complications during the procedure, and an endoscopic examination 3 months after the procedure revealed that the gastric varices had disappeared. To our knowledge, this is the first report on the application of plug-assisted retrograde transvenous obliteration-II using a newly designed IMPEDE vascular plug to avoid migration of the sclerosant.