Diagnosis of focal liver lesions from ultrasound images using a pretrained residual neural network.

OBJECTIVE: This study aims to develop a ResNet50-based deep learning model for focal liver lesion (FLL) classification in ultrasound images, comparing its performance with other models and prior research. METHODOLOGY: We retrospectively collected 581 ultrasound images from the Chulabhorn Hospital's HCC surveillance and screening project (2010-2018). The dataset comprised five classes: non-FLL, hepatic cyst (Cyst), hemangioma (HMG), focal fatty sparing (FFS), and hepatocellular carcinoma (HCC). We conducted 5-fold cross-validation after random dataset partitioning, enhancing training data with data augmentation. Our models used modified pre-trained ResNet50, GGN, ResNet18, and VGG16 architectures. Model performance, assessed via confusion matrices for sensitivity, specificity, and accuracy, was compared across models and with prior studies. RESULTS: ResNet50 outperformed other models, achieving a 5-fold cross-validation accuracy of 87 ± 2.2%. While VGG16 showed similar performance, it exhibited higher uncertainty. In the testing phase, the pretrained ResNet50 excelled in classifying non-FLL, cysts, and FFS. To compare with other research, ResNet50 surpassed the prior methods like two-layered feed-forward neural networks (FFNN) and CNN+ReLU in FLL diagnosis. CONCLUSION: ResNet50 exhibited good performance in FLL diagnosis, especially for HCC classification, suggesting its potential for developing computer-aided FLL diagnosis. However, further refinement is required for HCC and HMG classification in future studies.

Efficacy and Safety of a Dispersible Tablet of GPO-Deferasirox Monotherapy among Children with Transfusion-Dependent Thalassemia and Iron Overload.

The study aimed to determine efficacy and safety of generic deferasirox monotherapy. Deferasirox was administered in transfusion-induced iron overloaded thalassemia. Efficacy was defined as responders and nonresponders by ≤ 15 reduced serum ferritin from baseline. Adverse events were also monitored. Fifty-two patients with mainly Hb E/ß-thalassemia at the mean (SD) age of 8.7 (4.1) years, were enrolled. The mean (SD) daily transfusion iron load was 0.47 (0.1) mg/kg and maximum daily deferasirox was 35.0 (6.2) mg/kg. Altogether, 52, 40 and 18 patients completed the first, second and third years of study, respectively. The median baseline serum ferritin 2,383 ng/mL decreased to 1,478, 1,038 and 1,268 ng/mL at the end of first, second and third years, respectively, with overall response rate at 73.1% (38/52). Patients with baseline serum ferritin >2,500 ng/mL showed a change in serum ferritin higher than those ≤2,500 ng/mL starting from the 9th month of chelation. Adverse events were found in 5 of 52 patients (9.6%) including transaminitis (n = 2), one each of proteinuria, rash and proximal tubular dysfunction which resolved after transient stopping or decreasing the chelation dose. Generic deferasirox was effective and safe among pediatric patients with transfusion-induced iron overloaded thalassemia.

Axial loading during supine MRI for improved assessment of lumbar spine: comparison with standing MRI.

BACKGROUND: There are no studies comparing the morphologic changes of lumbar spines between supine axial-loaded and 90° standing magnetic resonance imaging (MRI) examinations of patients with spinal stenosis. PURPOSE: To determine whether axial-loaded MRI using a compression device demonstrated similar morphology of intervertebral disc, dural sac, and spinal curvature as those detected by 90° standing MRI in individuals with suspected spinal stenosis. MATERIAL AND METHODS: A total of 54 individuals suspected of having spinal stenosis underwent both axial-loaded and standing MRI studies. The outcome measures included seven radiologic parameters of the lumbar spine: measures of the intervertebral disc (i.e. cross-sectional area [DA], disc height [DH], and anteroposterior distance [DAP]), dural sac (cross-sectional area [DCSA]), spinal curvature (i.e. lumbar lordosis [LL] and L1-L3-L5 angle [LA]), and total lumbar spine height (LH). RESULTS: For agreement between the two methods, intraclass correlation coefficient (ICC) ≥ 0.8 was found for all seven radiologic parameters. Supine axial-loaded MRI underestimated LL but remained correlated (ICC = 0.83) with standing MRI. Minor differences between the two methods (≤5.0%) were observed in DA, DCSA, DAP, LA, and LH, while a major difference was observed in LL (8.1%). CONCLUSION: Using a compression device with the conventional supine MRI to simulate weight-bearing on the lumbar spine generated MRI morphology, which was strongly correlated with those from a standing MRI.

In vitro galactose-targeted study of RSPP050-loaded micelles against liver hepatocellular carcinoma.

Andrographolide is a group of diterpenoid lactone isolated from Andrographis paniculata (Burm. F.) NEES. One of the analogues is 19-O-triphenylmethylandrographolide (RSPP050) which possesses anticancer activity. In seeking to capitalise on the last property, we have investigated the in vitro tumour targeting capabilities and MRI imaging for hepatocellular carcinoma. In this study, we have designed galactose-targeted and non-targeted micelles comprised of poly(ethylene glycol)-b-poly(lactide) that enveloped RSPP050 as an anticancer agent and superparamagnetic iron oxide (SPIO) as a contrast agent. The targeting abilities were endeavored by examining the cellular uptake with MTT assay, fluorescence microscopy, Prussian blue staining, and in vitro MRI. Targeted SPIO micelles as a T2* contrast agent decreased the relative T2* MRI intensity at 3 h. Results revealed that galactose micelles displayed 10.91 ± 0.19% drug loading content, -37.17 ± 0.63 mV zeta potential, and these micelles at the concentration of 0.5 µg/ml exhibited higher cytotoxicity than non-targeted micelles and free RSPP050 after incubation for 24 h. Fluorescence microscopy and Prussian blue staining at 3 h demonstrated significant cellular uptake by HepG2 cells. Thus, anticancer activity of RSPP050 could be improved using galactose as a targeting ligand and theranostic function was achieved using SPIO.

Lumbosacral spinal compression device with the use of a cushion back support in supine MRI.

BACKGROUND: We hypothesized that axial-loaded magnetic resonance imaging (MRI), modified with the use of a cushion placed behind the lower back (i.e. BS-MRI method), would simulate the standing position more accurately than an axial-loaded MRI without a cushion back support (BS). PURPOSE: To determine whether the BS-MRI method demonstrated similar morphologies on intervertebral disc (IVD), dural sac, and spinal curvature as those detected on 90° standing MRIs in individuals with suspected spinal stenosis. MATERIAL AND METHODS: Twenty-five subjects underwent a BS-MRI, as well as axial-loaded and standing MRI studies. Outcome measures were four radiographic parameters of the lumbar spine: IVD height (DH); dural sac cross-sectional area (DCSA); and spinal curvature (i.e. lumbar lordosis [LL] and L1-L3-L5 angle [LA]). RESULTS: Major differences (>5%) between standing MRI and BS-MRI methods were observed in DCSA, DH, and LL. Major differences between standing and axial loaded MRIs were observed only in DCSA and LA. Although BS-MRIs demonstrate an image of the lumbar spine curvature (i.e. LA) which is closer to that when standing than axial-loaded MRIs, it is likely to overestimate both narrowing of dural sac and extent of LL. CONCLUSION: Using a compression device with a BS to simulate weight-bearing on the lumbar spine is not recommended due to: (i) overestimation of the narrowing of the dural sac and extent of LL; and (ii) underestimation of loss of disc height. Supine axial-loading produced DCSA and DH which were strongly correlated with those detected with standing MRIs. Exceptions were that LL and LA were underestimated.

Study of biodistribution and systemic toxicity of glucose functionalized SPIO/DOX micelles.

The present study examined the cytotoxicity and magnetic resonance imaging (MRI) distribution of cancer-targeted, MRI-visible polymeric micelles that encapsulate doxorubicin (DOX) and superparamagnetic iron oxide (SPIO) and are conjugated with glucose as a targeting ligand. In this study, the micelles were investigated the clinical potential of glucose-micelles, in vitro cytotoxicity assays of nonencapsulating or SPIO-and-DOX-coencapsulating micelles were performed on L929 mouse fibroblasts, and we found that glucose-micelles did not exert in vitro cytotoxic effects. Next, in vitro MRI detectability of glucose SPIO micelles was evaluated at the loaded SPIO content of 2.5% and 50%, and it was found that glucose-micelles can increase MRI relaxivity (r2*) at high SPIO loading. Furthermore, 50% SPIO micelles persisted in the blood circulation for up to 5 days (slow liver clearance) as determined by in vivo MRI. For in vivo toxicity evaluation, 50% SPIO/DOX micelles at a dose up to 18 (mg DOX)/(kg body weight) showed no impact on animal health according to clinical chemistry and clinical hematology laboratory testing. Altogether, these results indicate that glucose-micelles can serve as an effective and safe drug delivery system.

Upper Airway Areas, Volumes, and Linear Measurements Determined on Computed Tomography During Different Phases of Respiration Predict the Presence of Severe Obstructive Sleep Apnea.

PURPOSE: The objective of this study was to analyze the potential of using low-dose volumetric computed tomography (CT) during different phases of respiration for identifying patients likely to have severe obstructive sleep apnea (OSA), defined as a respiratory disturbance index (RDI) higher than 30. PATIENTS AND METHODS: A prospective study was undertaken at the Ramathibodi Hospital (Bangkok, Thailand). Patients with diagnosed OSA (N = 82) were recruited and separated into group 1 (RDI, ≤30; n = 36) and group 2 (RDI, >30; n = 46). The 2 groups were scanned by low-dose volumetric CT while they were 1) breathing quietly, 2) at the end of inspiration, and 3) at the end of expiration. Values for CT variables were obtained from linear measurements on lateral scout images during quiet breathing and from the upper airway area and volume measurements were obtained on axial cross-sections during different phases of respiration. All CT variables were compared between study groups. A logistic regression model was constructed to calculate a patient's likelihood of having an RDI higher than 30 and the predictive value of each variable and of the final model. RESULTS: The minimum cross-sectional area (MCA) measured at the end of inspiration (cutoff point, ≤0.33 cm2) was the most predictive variable for the identification of patients likely to have an RDI higher than 30 (adjusted odds ratio [OR] = 5.50; 95% confidence interval [CI], 1.76-17.20; sensitivity, 74%; specificity, 72%,), followed by the MCA measured at the end of expiration (cutoff point, ≤0.21 cm2; adjusted OR = 3.28; 95% CI, 1.05-10.24; sensitivity, 70%; specificity, 68%). CONCLUSION: CT scanning at the ends of inspiration and expiration helped identify patients with an RDI higher than 30 based on measurement of the MCA. Low-dose volumetric CT can be a useful tool to help the clinician rapidly identify patients with severe OSA and decide on the urgency to obtain a full-night polysomnographic study and to start treatment.

Computed Tomography Characterization and Comparison With Polysomnography for Obstructive Sleep Apnea Evaluation.

PURPOSE: We hypothesized that computed tomography (CT) combined with portable polysomnography (PSG) might better visualize anatomic data related to obstructive sleep apnea (OSA). The present study evaluated the CT findings during OSA and assessed their associations with the PSG data and patient characteristics. PATIENTS AND METHODS: We designed a prospective cross-sectional study of patients with OSA. The patients underwent scanning during the awake state and apneic episodes. Associations of the predictor variables (ie, PSG data, respiratory disturbance index [RDI]), patient characteristics (body mass index [BMI], neck circumference [NC], and waist circumference [WC]), and outcome variables (ie, CT findings during apneic episodes) were assessed using logistic regression analysis. The CT findings during apneic episodes were categorized regarding the level of obstruction, single level (retropalatal [RP] or retroglossal [RG]) or multilevel (mixed RP and RG), degree of obstruction (partial or complete), and pattern of collapse (complete concentric collapse [CCC] or other patterns). RESULTS: A total of 58 adult patients with OSA were scanned. The mean ± standard deviation for the RDI, BMI, NC, and WC were 41.6 ± 28.55, 27.80 ± 5.43 kg/m2, 38.3 ± 4.3 cm, and 93.8 ± 13.6 cm, respectively. No variables distinguished between the presence of single- and multilevel airway obstruction in the present study. A high RDI (≥30) was associated with the presence of complete obstruction and CCC (odds ratio 6.33, 95% confidence interval 1.55 to 25.90; and odds ratio 3.77, 95% confidence interval 1.02 to 13.91, respectively) compared with those with a lesser RDI. CONCLUSIONS: An increased RDI appears to be an important variable for predicting the presence of complete obstruction and CCC during OSA. Scanning during apneic episodes, using low-dose volumetric CT combined with portable PSG provided better anatomic and pathologic findings of OSA than did scans performed during the awake state.

Glucose-installed biodegradable polymeric micelles for cancer-targeted drug delivery system: synthesis, characterization and in vitro evaluation.

Glucose metabolism of cancer can be used as a strategy to target cancer cells which exhibit altered glycolytic rate. The facilitated glucose transporter (Glut) plays an important role in enhancement glycolytic rate resulting in increased glucose uptake into cancer cells. 18FGD-PET image is an example for using Glut as a targeting to diagnose the high glycolytic rate of tumor. Thus, Glut may be adapted to target cancer cells for drug delivery system. Herein, biodegradation polymeric micelles target cancer cells by Glut was fabricated. The amphiphilic block copolymer of poly(ethylene glycol)-block-poly(ε-caprolactone) (PEG-b-PCL) was synthesized where terminal group of the PEG chain was installed with glucose molecules. The 1H-NMR confirmed the existence of glucose moiety from two distinct peaks (5.2 and 4.7 ppm) of protons at anomeric carbon of glucose. Glucose-PEG-b-PCL spontaneously forms micelles in an aqueous solution. The size and zeta potential were 22 nm and -7 mv, respectively. Glucose-micelles have high stability, and no evidence of cytotoxicity was found after incubation for 7 days. Doxorubicin, used as a fluorescent probe, was loaded into glucose-micelles. The enhanced amount of doxorubicin as a result of glucose-micelles in PC-3, MCF-7 and HepG2 was evaluated by fluorescence microscopy and flow cytometer. Glucose molecules on the surface of micelles increased internalization and enhanced uptake of micelles via bypassing endocytosis pathway. These results show the use of glucose as a targeting ligand on the micelle surface to target cancer cells via Glut.

Effect of Iron Chelation Therapy on Glucose Metabolism in Non-Transfusion-Dependent Thalassaemia.

AIMS: To compare insulin sensitivity, ß-cell function and iron status biomarkers in non-transfusion-dependent thalassaemia (NTDT) with iron excess during pre- and post-iron chelation. METHODS: Subjects with NTDT, aged older than 10 years, with serum ferritin >300 ng/ml, were included. Iron chelation with deferasirox (10 mg/kg/day) was prescribed daily for 6 months. RESULTS: Ten patients with a median age of 17.4 years were enrolled. The comparison between pre- and post-chelation demonstrated significantly lower iron load: median serum ferritin (551.4 vs. 486.2 ng/ml, p = 0.047), median TIBC (211.5 vs. 233.5 µg/dl, p = 0.009) and median non-transferrin binding iron (5.5 vs. 1.4 µM, p = 0.005). All patients had a normal oral glucose tolerance test (OGTT) both pre- and post-chelation. However, fasting plasma glucose was significantly reduced after iron chelation (85.0 vs.79.5 mg/dl, p = 0.047). MRI revealed no significant changes of iron accumulation in the heart and liver after chelation, but there was a significantly lower iron load in the pancreas, assessed by higher T2* at post-chelation compared with pre-chelation (41.9 vs. 36.7 ms, p = 0.047). No adverse events were detected. CONCLUSIONS: A trend towards improving insulin sensitivity and ß-cell function as well as a reduced pancreatic iron load was observed following 6 months of iron chelation (TCTR20160523003).