Authorship：Lead author   Publishing type：Research paper (scientific journal)   International / domestic magazine：International journal  

OBJECTIVES: We aim to assess the early use of contrast-enhanced computed tomography (CECT) of patients with severe acute pancreatitis (SAP) using the computed tomography severity index (CTSI) in prognosis prediction. The CTSI combines quantification of pancreatic and extrapancreatic inflammation with the extent of pancreatic necrosis. METHODS: Post-hoc retrospective analysis of a large, multicentric database (44 institutions) of SAP patients in Japan. The area under the curve (AUC) of the CTSI for predicting mortality and the odds ratio (OR) of the extent of pancreatic inflammation and necrosis were calculated using multivariable analysis. RESULTS: In total, 1097 patients were included. The AUC of the CTSI for mortality was 0.65 (95 % confidence interval [CI:] [0.59-0.70]; p < 0.001). In multivariable analysis, necrosis 30-50 % and >50 % in low-enhanced pancreatic parenchyma (LEPP) was independently associated with a significant increase in mortality, with OR 2.04 and 95 % CI 1.01-4.12 (P < 0.05) and OR 3.88 and 95 % CI 2.04-7.40 (P < 0.001), respectively. However, the extent of pancreatic inflammation was not associated with mortality, regardless of severity. CONCLUSIONS: The degree of necrosis in LEPP assessed using early CECT of SAP was a better predictor of mortality than the extent of pancreatic inflammation.

DOI： 10.1016/j.pan.2024.07.001

PubMed

Publishing type：Research paper (scientific journal)   International / domestic magazine：International journal  

Numerous COVID-19 diagnostic imaging Artificial Intelligence (AI) studies exist. However, none of their models were of potential clinical use, primarily owing to methodological defects and the lack of implementation considerations for inference. In this study, all development processes of the deep-learning models are performed based on strict criteria of the "KAIZEN checklist", which is proposed based on previous AI development guidelines to overcome the deficiencies mentioned above. We develop and evaluate two binary-classification deep-learning models to triage COVID-19: a slice model examining a Computed Tomography (CT) slice to find COVID-19 lesions; a series model examining a series of CT images to find an infected patient. We collected 2,400,200 CT slices from twelve emergency centers in Japan. Area Under Curve (AUC) and accuracy were calculated for classification performance. The inference time of the system that includes these two models were measured. For validation data, the slice and series models recognized COVID-19 with AUCs and accuracies of 0.989 and 0.982, 95.9% and 93.0% respectively. For test data, the models' AUCs and accuracies were 0.958 and 0.953, 90.0% and 91.4% respectively. The average inference time per case was 2.83 s. Our deep-learning system realizes accuracy and inference speed high enough for practical use. The systems have already been implemented in four hospitals and eight are under progression. We released an application software and implementation code for free in a highly usable state to allow its use in Japan and globally.

DOI： 10.1038/s41598-024-52135-y

PubMed

Publishing type：Research paper (scientific journal)   International / domestic magazine：International journal  

BACKGROUND: Postneurosurgical meningitis (PNM) is a serious complication in neurocritical care patients, leading to clinical deterioration and worsening outcomes. Accurate diagnosis of PNM is often difficult due to the lack of definitive diagnostic criteria. This study investigates the potential utility of cerebrospinal fluid (CSF) presepsin (PSP), blood PSP, and the CSF/blood PSP ratio as adjunctive biomarkers for the diagnosis of PNM. METHODS: We conducted a single-center prospective observational study at Nara Prefecture General Medical Center in Nara, Japan, from April 2020 to March 2022. The postoperative neurosurgical patients with suspected PNM were included in the study and divided into PNM and non-PNM groups. We evaluated the sensitivity, specificity, area under curves (AUCs), positive predictive value (PPV), and negative predictive value (NPV) for the diagnosis of PNM with CSF PSP, blood PSP, and CSF/blood PSP ratio compared in the two groups. RESULTS: We screened 241 consecutive patients with postoperative neurosurgery. Diagnosis of PNM was suspected in 27 patients, and the clinical diagnosis was confirmed in nine patients. The results of CSF PSP (cutoff: 736 pg/mL) for the diagnosis of PNM were sensitivity 89%, specificity 78%, PPV 67%, NPV 93%, AUC 0.81 (95% confidence interval [CI], 0.60-1.00), blood PSP (cut-off: 264 pg/mL) was 56%, 78%, 56%, and 78%, 0.65 (95% CI, 0.42-0.88), and those of CSF/blood PSP ratio (cutoff: 3.45) was 89%, 67%, 57%, and 92%, 0.83 (95% CI, 0.65-1.00). CONCLUSION: Elevated CSF PSP and CSF/blood PSP ratio may be associated with PNM and could serve as valuable adjunctive biomarkers for improving diagnostic accuracy.

DOI： 10.25259/SNI_903_2023

PubMed