Artificial intelligence model predicting postoperative pain using facial expressions: a pilot study.

PURPOSE: This study aimed to assess whether an artificial intelligence model based on facial expressions can accurately predict significant postoperative pain. METHODS: A total of 155 facial expressions from patients who underwent gastric cancer surgery were analyzed to extract facial action units (AUs), gaze, landmarks, and positions. These features were used to construct various machine learning (ML) models, designed to predict significant postoperative pain intensity (NRS ≥ 7) from less significant pain (NRS < 7). Significant AUs predictive of NRS ≥ 7 were determined and compared to AUs known to be associated with pain in awake patients. The area under the receiver operating characteristic curves (AUROCs) of the ML models was calculated and compared using DeLong's test. RESULTS: AU17 (chin raising) and AU20 (lip stretching) were found to be associated with NRS ≥ 7 (both P ≤ 0.004). AUs known to be associated with pain in awake patients did not show an association with pain in postoperative patients. An ML model based on AU17 and AU20 demonstrated an AUROC of 0.62 for NRS ≥ 7, which was inferior to a model based on all AUs (AUROC = 0.81, P = 0.006). Among facial features, head position and facial landmarks proved to be better predictors of NRS ≥ 7 (AUROC, 0.85-0.96) than AUs. A merged ML model that utilized gaze and eye landmarks, as well as head position and facial landmarks, exhibited the best performance (AUROC, 0.90) in predicting significant postoperative pain. CONCLUSION: ML models using facial expressions can accurately predict the presence of significant postoperative pain and have the potential to screen patients in need of rescue analgesia. TRIAL REGISTRATION NUMBER: This study was registered at ClinicalTrials.gov (NCT05477303; date: June 17, 2022).

Modified Thoracoabdominal Nerves Block Through Perichondral Approach: A Systematic Review and Meta-analysis.

PURPOSE: This systematic review and meta-analysis aimed to investigate the postoperative analgesic efficacy and safety of the modified thoracoabdominal nerve block through the perichondral approach (M-TAPA) in abdominal surgeries. DESIGN: Systematic review and meta-analysis. METHODS: We searched electronic databases to identify relevant studies comparing M-TAPA with conventional analgesic techniques. The primary outcome was the requirement for rescue analgesia at 12 and 24 hours postsurgery. Secondary outcomes included the 11-point numerical rating scale pain scores at 0, 1, 2, 4, 6, 8, 12, and 24 hours following surgery, global quality of recovery scores, and postoperative adverse events. FINDINGS: Five randomized controlled trials involving 308 patients were analyzed. M-TAPA showed no significant difference in the requirement for rescue analgesia at 12 hours (relative risk [RR]: 0.87; 95% confidence interval [CI]: 0.62, 1.22; P = .424; I2 = 40.7%; Ph = .185) and 24 hours (RR: 0.67; 95% CI: 0.22, 1.99; P = .252; I2 = 90.3%; Ph < .001) postsurgery compared to non-M-TAPA. No significant differences in numerical rating scale pain scores or global quality of recovery scores were found between the two groups (all P < .05). However, M-TAPA was associated with a lower occurrence of nausea (RR: 0.37; 95% CI: 0.22, 0.68; P < .001; I2 = 0%; Ph = .834), vomiting (RR: 0.32; 95% CI: 0.17, 0.62; P < .001; I2 = 0%; Ph = .884), and itching (RR: 0.38; 95% CI: 0.21, 0.70; P = .002; I2 = 0%; Ph = .826). CONCLUSIONS: There was no significant difference in analgesic efficacy and safety between M-TAPA and non-M-TAPA techniques.

Steatotic hepatocellular carcinoma: Association of MRI findings to underlying liver disease and clinicopathological characteristics.

BACKGROUND & AIMS: Fatty change is commonly observed in hepatocellular carcinoma (HCC); however, the characteristics of steatotic and steatohepatitic HCCs are not well understood. METHODS: This retrospective study included patients with HCCs who underwent resection between January 2014 and December 2019 to evaluate clinicopathological and magnetic resonance imaging features. Tumours were categorized as magnetic resonance imaging-steatotic, pathology-steatotic and steatohepatitic HCCs and were defined as HCCs with ≥50% steatosis on in-and-oppose phase images, ≥34% tumour cells with lipid droplets and ≥50% tumour areas with steatohepatitic features on light microscopy respectively. RESULTS: Of 465 HCCs, 38 (8%), 23 (5%) and 15 (3%) were diagnosed as magnetic resonance imaging-steatotic, pathology-steatotic and steatohepatitic HCCs respectively. These HCC variants were less likely to be associated with hepatitis B virus infections than with type 2 diabetes mellitus, metabolic syndrome, non-tumour liver steatosis and steatohepatitis. Moreover, microvascular invasion was less likely to be associated with them than either tumour size or differentiation. Type 2 diabetes and non-tumour steatosis were independent risk factors for magnetic resonance imaging-steatotic HCCs. Pathology-steatotic HCCs and steatohepatitic HCCs were significantly associated with magnetic resonance imaging-steatotic HCCs. A targetoid appearance in the transitional or hepatobiliary phase was also more prevalent in steatohepatitic-HCCs than in non-steatohepatitic-HCCs. When magnetic resonance imaging-steatotic HCCs were combined with one or more ancillary features, the sensitivity and specificity were 60% and 97% respectively. CONCLUSION: Underlying fatty liver disease and metabolic syndrome are strongly associated with both steatotic and steatohepatitic HCCs. Clinicoradiological characteristics help identify steatohepatitic HCC with high specificity.

Noninvasive evaluation of liver fibrosis: comparison of the stretched exponential diffusion-weighted model to other diffusion-weighted MRI models and transient elastography.

OBJECTIVES: To compare the diagnostic performance of the stretched exponential model to those of other DWI models and transient elastography (TE) and to evaluate the influence of confounding factors on the staging of liver fibrosis. METHODS: This retrospective study included 78 consecutive patients who underwent both DWI and TE. The distributed diffusion coefficient (DDC) and intravoxel heterogeneity index (α) from the stretched exponential model, apparent diffusion coefficient (ADC), perfusion fraction (f), pseudodiffusion coefficient (Dp), true diffusion coefficient (Dt), and TE were obtained. Associations between imaging parameters and pathological fibrosis, inflammation, and steatosis were evaluated using Spearman's correlation and multiple regression analysis. Diagnostic accuracy of parameters for fibrosis staging was assessed via the Obuchowski measures. RESULTS: DDC was the only parameter to differ between F0-1 and F2-3 (p < 0.001) and between F2-3 and F4 (p = 0.013). DDC showed significant correlation with fibrosis (p < 0.001) and inflammation (p = 0.001), but not with steatosis (p = 0.619), and was independently associated with only fibrosis in multiple regression analysis (ß = - 0.114, p < 0.001). ADC, Dp, and Dt showed a significant correlation with steatosis (ps ≤ 0.038). DDC showed the highest diagnostic performance for liver fibrosis (0.717; 95% confidence interval, 0.653-0.765) followed by TE (0.681, 0.623-0.733) without a significant difference between DDC and TE (p > 0.999). CONCLUSIONS: DDC from the stretched exponential model is the most accurate DWI parameter with no confounding effect from steatosis and with overall similar diagnostic performance to TE. KEY POINTS: • The distributed diffusion coefficient (DDC) from the stretched exponential model is the most accurate DWI parameter for staging liver fibrosis. • DDC and transient elastography have similar good diagnostic performance for evaluating liver fibrosis. • The stretched exponential DWI model has no confounding effect by steatosis, unlike other DWI models.

Genetic variations in MicroRNA genes and cancer risk: A field synopsis and meta-analysis.

BACKGROUND: Cancer risk has been associated with certain gene variations in microRNA (miRNA), but conflicting evidence warrants re-assessing of significant results in meta-analyses. We summarized published meta-analyses that assess the associations between miRNA polymorphism and cancers to show the validity of the findings. METHOD: We searched PubMed and investigated the results of meta-analyses published through November 2018. We re-assessed the results based on false-positive report probability (FPRP) to test the noteworthiness of the associations. RESULTS: Sixty-eight miRNA polymorphisms in 45 meta-analyses associated with cancer were included. Four (7.4%) and sixteen (25.0%) single nucleotide polymorphisms (SNPs) were noteworthy (FPRP < 0.2) at a prior probability of 0.001 for interesting candidate genes and a statistical power to detect an odds ratio (OR) of 1.1 and 1.5, respectively. The four miRNA SNPs noteworthy at an OR of 1.1 were as follows: miR-146a/rs2910164 Cvs.G; miR-27a/rs895819 Cvs.T; miR-423/rs6505162 Cvs.A; and miR-605/rs2043556 Cvs.T. The 16 SNPs noteworthy at an OR of 1.5 include the four genotype comparisons at an OR of 1.1, and the additional 12 genotype comparisons were as follows: miR-196a2/rs11614913 Tvs.C; miR-27a/rs895819 GGvs.AA + AG; miR-196a2/rs11614913 C vs.T; miR-146a/rs2910164 Gvs.C; miR-196a2/rs11614913 Tvs.C; miR-146a/rs2910164 Cvs.G; miR-499/rs3746444 homozygous model; miR-146a/rs2910164 CCvs.GG + GC; miR-499/rs3746444 TCvs.TT; miR-499/rs3746444 GAvs.AA; miR-146a/rs2910164 CCvs.GG; and miR-499/rs3746444 Gvs.A. No association was noteworthy at a prior probability of 0.000001. CONCLUSION: Out of 68 published associations of miRNA polymorphisms with cancer, sixteen have shown noteworthiness in our re-assessing meta-analysis. Our findings summarize the results of meta-analyses on the association of cancer with SNPs and underline the importance of interpreting results with caution.

Is there association between statin usage and contrast-associated acute kidney injury after intravenous administration of iodine-based contrast media in enhanced computed tomography?

OBJECTIVES: Contrast-induced acute kidney injury (CI-AKI) is one of the leading causes of new-onset renal failure in hospitalized patients. Statin has been evaluated for its protective effect against CI-AKI but rarely in patients receiving intravenous (IV) administrations of iodine-based contrast media for enhanced computed tomography (CT). METHODS: In total, 12,371 patients who underwent contrast-enhanced abdominopelvic CT were retrospectively reviewed and stratified into statin users and statin nonusers. Subgroup analyses comparing high-intensity statins with low- to moderate-intensity statins were conducted within statin users and similar comparisons were performed within statin users stratified based on baseline eGFR. RESULTS: Overall, CI-AKI events did not occur less in statin users compared with non-statin users (p = 0.342). Within statin users, CI-AKI events did not decrease in high-intensity statin users compared with low- to moderate-intensity statin users (p = 0.355). Moreover, no significant difference in CI-AKI events was found between high-intensity statin users and low- to moderate-intensity statin users even after stratifying the patients with baseline eGFR. CONCLUSIONS: Collectively, statin was not significantly associated with the risk of CI-AKI events in patients undergoing contrast-enhanced abdominopelvic CT and high-intensity statins did not show significant association with CI-AKI over low- to moderate-intensity statins in the subgroup analysis. KEY POINTS: • Statin is not associated with risk of CI-AKI events in patients undergoing intravenous administration of contrast-enhanced CT. • CI-AKI incidence among high-intensity statin users was not significantly different from that of low- to moderate-intensity statin users.

Posterior nutcracker syndrome - a systematic review.

Posterior nutcracker syndrome (PNCS) is the entrapment of the left renal vein between the aorta and the vertebral column. Although uncommon, it is still an important diagnosis due to the high morbidity associated with the risk of secondary anaemia from haematuria, from long-term left renal vein hypertension, vascular thrombosis, and even blood clots in the urinary system. A literature search of PubMed and EMBASE databases was performed and 27 publications containing 27 cases were included for the final analysis. The following frequency of clinical signs and symptoms was noted: twenty-five patients had haematuria, 13 patients had flank pain, and two had hypertension. Overall, male-female distribution was balanced and there were more adult than paediatric (age < 18 years) patients. All symptoms of patients with conservative treatment were either well-controlled or under spontaneous resolution. Conservative management instead of surgical treatment should be preferred in most cases. Taken together, despite the low incidence of PNCS, its recognition and management are highly important. This systematic study explores the evidence base for conservative and medical options.

Delta neutrophil index (DNI) as a novel diagnostic and prognostic marker of infection: a systematic review and meta-analysis.

OBJECTIVE: Delta neutrophil index (DNI) representing the number of immature granulocytes is an emerging marker used in diagnosis of infections and prediction of mortality in infected patients. The present study evaluated the diagnostic accuracy of DNI as a predictive and prognostic factor in infected patients. METHODS: We performed a PubMed search on January 1st, 2017 and identified studies that evaluated DNI as either a predictive or prognostic factor in infected patients. Studies with appropriate information to construct 2 × 2 contingency tables were extracted. We calculated pooled sensitivity and specificity. Meta-analysis of the multivariate logistic regression data set was performed to assess whether DNI functions as an independent factor. RESULTS: Overall, 12 articles fulfilled the inclusion criteria and a total of 499 cases and 9549 controls were examined. As a predictive factor of infection, DNI's pooled sensitivity was 0.67 (95% CI 0.62-0.71, I 2 = 86.0%) and pooled specificity was 0.94 (95% CI 0.94-0.95, I 2 = 92.8%). Area under the receiver operating characteristics (ROC) curve was 0.89. As a prognostic factor for death in infected patients, DNI's pooled sensitivity was 0.70 (95% CI 0.56-0.81, I 2 = 0.0%) and pooled specificity was 0.78 (95% CI 0.73-0.83, I 2 = 26.6%). Area under the ROC curve was 0.84. Meta-analysis of the multivariate logistic regression data set showed insignificant results. CONCLUSIONS: DNI is a potentially useful diagnostic tool and predicts mortality among infected patients and should be more widely used in the clinical practice.

A Multi-RNAi Microsponge Platform for Simultaneous Controlled Delivery of Multiple Small Interfering RNAs.

Packaging multiple small interfering RNA (siRNA) molecules into nanostructures at precisely defined ratios is a powerful delivery strategy for effective RNA interference (RNAi) therapy. We present a novel RNA nanotechnology based approach to produce multiple components of polymerized siRNA molecules that are simultaneously self-assembled and densely packaged into composite sponge-like porous microstructures (Multi-RNAi-MSs) by rolling circle transcription. The Multi-RNAi-MSs were designed to contain a combination of multiple polymeric siRNA molecules with precisely controlled stoichiometry within a singular microstructure by manipulating the types and ratios of the circular DNA templates. The Multi-RNAi-MSs were converted into nanosized complexes by polyelectrolyte condensation to manipulate their physicochemical properties (size, shape, and surface charge) for favorable delivery, while maintaining the multifunctional properties of the siRNAs for combined therapeutic effects. These Multi-RNAi-MS systems have great potential in RNAi-mediated biomedical applications, for example, for the treatment of cancer, genetic disorders, and viral infections.

Mesial temporal atrophy in preoperative MRI rather than steep Trendelenburg position is associated with postoperative delirium in patients undergoing a major urologic surgery.

PURPOSE: To investigate whether steep Trendelenburg in a major urologic surgery is associated with postoperative delirium, and to examine other potential clinical and radiologic factors predictive of postoperative delirium. METHODS: 182 patients who received a major urologic surgery and underwent a 3.0-T brain MRI scan within 1 year prior to the date of surgery were retrospectively enrolled. Preoperative brain MRIs were used to analyze features related to small vessel disease burden and mesial temporal atrophy. Presence of a significant mesial temporal atrophy was defined as Scheltens' scale ≥ 2. Patients' clinico-demographic data and MRI features were used to identify significant predictors of postoperative delirium using the logistic regression analysis. Independent predictors found significant in the univariate analysis were further evaluated in the multivariate analysis. RESULTS: Incidence of postoperative delirium was 6.0%. Patients with postoperative delirium had lower body mass index (21.3 vs. 25.0 kg/m2, P = 0.003), prolonged duration of anesthesia (362.7 vs. 224.7 min, P < 0.001) and surgery (302.2 vs. 174.5 min, P < 0.001), and had more significant mesial temporal atrophy (64% vs. 30%, P = 0.046). In the univariate analysis, female sex, type of surgery (radical prostatectomy over cystectomy), prolonged duration of anesthesia (≥ 6 h), and presence of a significant mesial temporal atrophy were significant predictors (all P-values < 0.050), but only the presence of significant mesial temporal atrophy was significant in the multivariate analysis [odds ratio (OR), 3.69; 95% CI 0.99-13.75; P = 0.046]. CONCLUSION: Steep Trendelenburg was not associated with postoperative delirium. Significant mesial temporal atrophy (Scheltens' scale ≥ 2) in preoperative brain MRI was predictive of postoperative delirium. TRIAL REGISTRATION: Not applicable.

Predicting stroke volume variation using central venous pressure waveform: a deep learning approach.

Objective. This study evaluated the predictive performance of a deep learning approach to predict stroke volume variation (SVV) from central venous pressure (CVP) waveforms.Approach. Long short-term memory (LSTM) and the feed-forward neural network were sequenced to predict SVV using CVP waveforms obtained from the VitalDB database, an open-source registry. The input for the LSTM consisted of 10 s CVP waveforms sampled at 2 s intervals throughout the anesthesia duration. Inputs of the feed-forward network were the outputs of LSTM and demographic data such as age, sex, weight, and height. The final output of the feed-forward network was the SVV. The performance of SVV predicted by the deep learning model was compared to SVV estimated derived from arterial pulse waveform analysis using a commercialized model, EV1000.Main results. The model hyperparameters consisted of 12 memory cells in the LSTM layer and 32 nodes in the hidden layer of the feed-forward network. A total of 224 cases comprising 1717 978 CVP waveforms and EV1000/SVV data were used to construct and test the deep learning models. The concordance correlation coefficient between estimated SVV from the deep learning model were 0.993 (95% confidence interval, 0.992-0.993) for SVV measured by EV1000.Significance. Using a deep learning approach, CVP waveforms can accurately approximate SVV values close to those estimated using commercial arterial pulse waveform analysis.

Postoperative analgesic effects of the quadratus lumborum block in pediatric patients: a systematic review and meta-analysis.

Background: This study assessed the postoperative analgesic efficacy and safety of the quadratus lumborum block (QLB) in pediatric patients. Methods: Electronic databases were searched for studies comparing the QLB to conventional analgesic techniques in pediatric patients. The primary outcome was the need for rescue analgesia 12 and 24 hours after surgery. Secondary outcomes covered the Face-Legs-Activity-Cry-Consolability Scale (FLACC) scores at various time points; parental satisfaction; time to the first rescue analgesia; hospitalization time; block execution time; block failure rates, and adverse events. Results: Sixteen randomized controlled trials were analyzed involving 1,061 patients. The QLB significantly reduced the need for rescue analgesia both at 12 and 24 hours after surgery (12 hours, relative risk [RR]: 0.45; 95% confidence interval [CI]: 0.01, 0.88; 24 hours, RR: 0.51; 95% CI: 0.31, 0.70). In case of 24 hours after surgery, type 1 QLB significantly reduced the need for rescue analgesia (RR: 0.56; 95% CI: 0.36, 0.76). The QLB also exhibited lower FLACC scores at 1 hour (standardized mean difference [SMD]: -0.87; 95% CI: -1.56, -0.18) and 6 hours (SMD: -1.27; 95% CI: -2.33, -0.21) following surgery when compared to non-QLB. Among QLBs, type 2 QLB significantly extended the time until the first rescue analgesia (SMD: 1.25; 95% CI: 0.84, 1.67). No significant differences were observed in terms of parental satisfaction, hospitalization time, block execution time, block failure, or adverse events between QLB and non-QLB groups. Conclusions: The QLB provides non-inferior analgesic efficacy and safety to conventional methods in pediatric patients.

Peripheral zone thickness in preoperative MRI is predictive of Trifecta achievement after Holmium laser enucleation of the prostate (HoLEP).

PURPOSE: To investigate various anatomical features of the prostate using preoperative MRI and patients' clinical factors to identify predictors of successful Holmium:YAG laser enucleation of the prostate (HoLEP). METHODS: 71 patients who had received HoLEP and undergone a 3.0-T prostate MRI scan within 6 months before surgery were retrospectively enrolled. MRI features (e.g., total prostate and transitional zone volume, peripheral zone thickness [PZT], BPH patterns, prostatic urethral angle, intravesical prostatic protrusion, etc.) and clinical data (e.g., age, body mass index, surgical technique, etc.) were analyzed using univariable and multivariable logistic regression to identify predictors of successful HoLEP. Successful HoLEP was defined as achieving the Trifecta, characterized by the contemporary absence of postoperative complications within 3 months, a 3-month postoperative maximum flow rate (Qmax) > 15 mL/s, and no urinary incontinence at 3 months postoperatively. RESULTS: Trifecta achievement at 3 months post-surgery was observed in 37 (52%) patients. Patients with Trifecta achievement exhibited a lower preoperative IPSS-quality of life score (QoL) (4.1 vs. 4.5, P = 0.016) and a thinner preoperative peripheral zone thickness (PZT) on MRI (7.9 vs.10.3 mm, P < 0.001). In the multivariable regression analysis, a preoperative IPSS-QoL score < 5 (OR 3.98; 95% CI, 1.21-13.07; P = 0.017) and PZT < 9 mm (OR 11.51; 95% CI, 3.51-37.74; P < 0.001) were significant predictors of Trifecta achievement after HoLEP. CONCLUSIONS: Alongside the preoperative QoL score, PZT measurement in prostate MRI can serve as an objective predictor of successful HoLEP. Our results underscore an additional utility of prostate MRI beyond its role in excluding concurrent prostate cancer.

Tumor necrosis in magnetic resonance imaging predicts urothelial carcinoma with squamous differentiation in muscle-invasive bladder carcinoma.

PURPOSE: This study investigated radiologic features on preoperative MRI to differentiate urothelial carcinoma with squamous differentiation (UCSD) from conventional urothelial carcinoma (UC) in muscle-invasive bladder carcinoma. METHODS: Ninety-nine patients who underwent radical cystectomy and a preoperative bladder MRI scan within three months before surgery were retrospectively enrolled. Various MRI features, including tumor length, location, multiplicity, long-to-short axis ratio, morphology, radiologic stage, and degree of severe necrosis, were analyzed. Univariable and multivariable logistic regression analyses were performed to identify MRI features predictive of UCSD. The diagnostic performance of a significant MRI feature was assessed using 5-fold cross-validation. RESULTS: Among the MRI features, significant radiologic findings associated with UCSD in the univariable analysis included heterogeneous tumor signal intensity in T2-weighted images (odds ratio [OR], 3.365; 95% confidence interval [CI], 1.213-9.986; P = 0.022) and contrast-enhanced T1-weighted images (OR, 4.428; 95% CI, 1.519-12.730; P = 0.007), as well as marked (≥ 50%) severe necrosis (OR, 17.100; 95% CI, 4.699-73.563; P < 0.001). In the multivariable analysis, marked (≥ 50%) severe necrosis (odds ratio [OR], 13.755; 95% confidence interval [CI], 2.796-89.118; P = 0.004) was a significant predictor of UCSD. Marked (≥ 50%) severe necrosis showed a high specificity of 95.0% with a precision of 65.0% for diagnosing UCSD based on 5-fold cross-validation. CONCLUSION: Preoperative bladder MRI revealing marked severe necrosis may be indicative of UCSD and can assist in distinguishing it from conventional UC.

Machine learning model of facial expression outperforms models using analgesia nociception index and vital signs to predict postoperative pain intensity: a pilot study.

BACKGROUND: Few studies have evaluated the use of automated artificial intelligence (AI)-based pain recognition in postoperative settings or the correlation with pain intensity. In this study, various machine learning (ML)-based models using facial expressions, the analgesia nociception index (ANI), and vital signs were developed to predict postoperative pain intensity, and their performances for predicting severe postoperative pain were compared. METHODS: In total, 155 facial expressions from patients who underwent gastrectomy were recorded postoperatively; one blinded anesthesiologist simultaneously recorded the ANI score, vital signs, and patient self-assessed pain intensity based on the 11-point numerical rating scale (NRS). The ML models' area under the receiver operating characteristic curves (AUROCs) were calculated and compared using DeLong's test. RESULTS: ML models were constructed using facial expressions, ANI, vital signs, and different combinations of the three datasets. The ML model constructed using facial expressions best predicted an NRS ≥ 7 (AUROC 0.93) followed by the ML model combining facial expressions and vital signs (AUROC 0.84) in the test-set. ML models constructed using combined physiological signals (vital signs, ANI) performed better than models based on individual parameters for predicting NRS ≥ 7, although the AUROCs were inferior to those of the ML model based on facial expressions (all P < 0.050). Among these parameters, absolute and relative ANI had the worst AUROCs (0.69 and 0.68, respectively) for predicting NRS ≥ 7. CONCLUSIONS: The ML model constructed using facial expressions best predicted severe postoperative pain (NRS ≥ 7) and outperformed models constructed from physiological signals.

Differentiation between hepatic angiomyolipoma and hepatocellular carcinoma in individuals who are not at-risk for hepatocellular carcinoma.

PURPOSE: To develop a practical methodfor differentiating hepatocellular carcinoma (HCC) from angiomyolipoma (AML) in individuals who are not at-risk for HCC. METHOD: We retrospectively enrolled consecutive patients who underwent gadoxetic acid-enhanced liver magnetic resonance imaging (MRI) and pathological confirmation between January 2008 and April 2022. Patients who underwent prior treatment, those with multiple lesions, or those at-risk for HCC were excluded. The training cohort included patients with pathological confirmation between 2008 and 2019, whereas the validation cohort included the remaining cases. Independent reviews of the MRI were performed by two reviewers. Using the clinical and MRI findings, we developed AML-HCC score using Firth's logistic regression in the training cohort, and the diagnostic performance was validated in the validation cohort. RESULTS: Of the 206 patients, 156 were assigned to the training cohort (25 and 131 patients with AML and HCC, respectively) and 50 were assigned to the validation cohort (4 and 46 patients with AML and HCC, respectively). The AML-HCC score was defined as the sum of female (score 1), early draining vein (score 2), T2 homogeneity (score 1), necrosis or severe ischaemia (score -2), and HBP hyperintensity to spleen (score -1). When the AML-HCC score was ≥1, the sensitivity and specificity were 80% and 95% for the training cohort and 100% and 80% for the validation cohort, respectively. CONCLUSIONS: We developed and validated an AML-HCC score to differentiate between AML and HCC in individuals who are not at-risk for HCC, and our model demonstrated good diagnostic performance.

A deep learning algorithm to quantify AVF stenosis and predict 6-month primary patency: a pilot study.

Background: A deep convolutional neural network (DCNN) model that predicts the degree of arteriovenous fistula (AVF) stenosis and 6-month primary patency (PP) based on AVF shunt sounds was developed, and was compared with various machine learning (ML) models trained on patients' clinical data. Methods: Forty dysfunctional AVF patients were recruited prospectively, and AVF shunt sounds were recorded before and after percutaneous transluminal angioplasty using a wireless stethoscope. The audio files were converted to melspectrograms to predict the degree of AVF stenosis and 6-month PP. The diagnostic performance of the melspectrogram-based DCNN model (ResNet50) was compared with that of other ML models [i.e. logistic regression (LR), decision tree (DT) and support vector machine (SVM)], as well as the DCNN model (ResNet50) trained on patients' clinical data. Results: Melspectrograms qualitatively reflected the degree of AVF stenosis by exhibiting a greater amplitude at mid-to-high frequency in the systolic phase with a more severe degree of stenosis, corresponding to a high-pitched bruit. The proposed melspectrogram-based DCNN model successfully predicted the degree of AVF stenosis. In predicting the 6-month PP, the area under the receiver operating characteristic curve of the melspectrogram-based DCNN model (ResNet50) (≥0.870) outperformed that of various ML models based on clinical data (LR, 0.783; DT, 0.766; SVM, 0.733) and that of the spiral-matrix DCNN model (0.828). Conclusion: The proposed melspectrogram-based DCNN model successfully predicted the degree of AVF stenosis and outperformed ML-based clinical models in predicting 6-month PP.

Feasibility of Deep Learning-Based Analysis of Auscultation for Screening Significant Stenosis of Native Arteriovenous Fistula for Hemodialysis Requiring Angioplasty.

OBJECTIVE: To investigate the feasibility of using a deep learning-based analysis of auscultation data to predict significant stenosis of arteriovenous fistulas (AVF) in patients undergoing hemodialysis requiring percutaneous transluminal angioplasty (PTA). MATERIALS AND METHODS: Forty patients (24 male and 16 female; median age, 62.5 years) with dysfunctional native AVF were prospectively recruited. Digital sounds from the AVF shunt were recorded using a wireless electronic stethoscope before (pre-PTA) and after PTA (post-PTA), and the audio files were subsequently converted to mel spectrograms, which were used to construct various deep convolutional neural network (DCNN) models (DenseNet201, EfficientNetB5, and ResNet50). The performance of these models for diagnosing ≥ 50% AVF stenosis was assessed and compared. The ground truth for the presence of ≥ 50% AVF stenosis was obtained using digital subtraction angiography. Gradient-weighted class activation mapping (Grad-CAM) was used to produce visual explanations for DCNN model decisions. RESULTS: Eighty audio files were obtained from the 40 recruited patients and pooled for the study. Mel spectrograms of "pre-PTA" shunt sounds showed patterns corresponding to abnormal high-pitched bruits with systolic accentuation observed in patients with stenotic AVF. The ResNet50 and EfficientNetB5 models yielded an area under the receiver operating characteristic curve of 0.99 and 0.98, respectively, at optimized epochs for predicting ≥ 50% AVF stenosis. However, Grad-CAM heatmaps revealed that only ResNet50 highlighted areas relevant to AVF stenosis in the mel spectrogram. CONCLUSION: Mel spectrogram-based DCNN models, particularly ResNet50, successfully predicted the presence of significant AVF stenosis requiring PTA in this feasibility study and may potentially be used in AVF surveillance.

Update on Pathologic and Radiologic Diagnosis of Combined Hepatocellular-Cholangiocarcinoma.

Combined hepatocellular-cholangiocarcinoma (cHCC-CCA) is a malignant primary liver carcinoma characterized by the unequivocal presence of both hepatocytic and cholangiocytic differentiation within the same tumor. Recent research has highlighted that cHCC-CCAs are more heterogeneous than previously expected. In the updated consensus terminology and WHO 2019 classification, "classical type" and "subtypes with stem-cell features" of the WHO 2010 classification are no longer recommended. Instead, it is recommended that the presence and percentages of various histopathologic components and stem-cell features be mentioned in the pathologic report. The new terminology and classification enable the exchange of clearer and more objective information about cHCC-CCAs, facilitating multi-center and multi-national research. However, there are limitations to the diagnosis of cHCC-CCA by imaging and biopsy. cHCC-CCAs showing typical imaging findings of HCC could be misdiagnosed as HCC and subjected to inappropriate treatment, if other clinical findings are not sufficiently considered. cHCC-CCAs showing at least one of the CCA-like imaging features or unusual clinical features should be subjected to biopsy. There may be a sampling error for the biopsy diagnosis of cHCC-CCA. An optimized diagnostic algorithm integrating clinical, radiological, and histopathologic information of biopsy is required to resolve these diagnostic pitfalls.

Hepatobiliary phase signal intensity: A potential method of diagnosing HCC with atypical imaging features among LR-M observations.

Herein, we assessed whether hepatobiliary phase (HBP) signal intensity (SI) can be used to differentiate HCC and non-HCC malignancies within LR-M observations. 106 LR-M patients based on LI-RADS v2018 who underwent gadoxetate-disodium magnetic resonance imaging and surgery from January 2009 to December 2018 were included. SI of LR-M observation on HBP was analyzed by two radiologists and categorized into dark, low and iso-to-high groups. Tumor was classified as dark when more than 50% of tumor showed hypointensity compared to spleen, as low when more than 50% of tumor showed hyperintensity compared to spleen but hypointensity compared to liver parenchyma, and as iso-to-high if there was even a focal iso-intensity or hyperintensity compared to liver parenchyma. Analysis of clinicopathological factors and association between imaging and histology was performed. Out of 106 LR-M, 42 (40%) were showed dark, 61 (58%) showed low, and 3 (3%) showed iso-to-high SI in HBP. Three iso-to-high SI LR-M were HCCs (P = 0.060) and their major histologic differentiation was Edmondson grade 1 (P = 0.001). 43 out of 61 (71%) low SI LR-M were iCCA or cHCC-CCA (P = 0.002). Inter-reader agreement of HBP SI classification was excellent, with a kappa coefficient of 0.872. LR-M with iso-to-high SI in HBP is prone to being HCC while LR-M with low SI in HBP is prone to being tumor with fibrous stroma such as iCCA and cHCC-CCA. Classification of LR-M based on HBP SI may be a helpful method of differentiating HCC from non-HCC malignancies.