MyMedR

Displaying all 7 publications

Abstract:

Sort:

Current status and future directions of explainable artificial intelligence in medical imaging

Saw SN, Yan YY, Ng KH

Eur J Radiol, 2025 Feb;183:111884.
PMID: 39667118 DOI: 10.1016/j.ejrad.2024.111884

The inherent "black box" nature of AI algorithms presents a substantial barrier to the widespread adoption of the technology in clinical settings, leading to a lack of trust among users. This review begins by examining the foundational stages involved in the interpretation of medical images by radiologists and clinicians, encompassing both type 1 (fast thinking - ability of the brain to think and act intuitively) and type 2 (slow analytical - slow analytical, laborious approach to decision-making) decision-making processes. The discussion then delves into current Explainable AI (XAI) approaches, exploring both inherent and post-hoc explainability for medical imaging applications and highlighting the milestones achieved. XAI in medicine refers to AI system designed to provide transparent, interpretable, and understandable reasoning behind AI predictions or decisions. Additionally, the paper showcases some commercial AI medical systems that offer explanations through features such as heatmaps. Opportunities, challenges and potential avenues for advancing the field are also addressed. In conclusion, the review observes that state-of-the-art XAI methods are not mature enough for implementation, as the explanations they provide are challenging for medical experts to comprehend. Deeper understanding of the cognitive mechanisms by medical professionals is important in aiming to develop more interpretable XAI methods.
Machine learning improves early prediction of small-for-gestational-age births and reveals nuchal fold thickness as unexpected predictor

Saw SN, Biswas A, Mattar CNZ, Lee HK, Yap CH

Prenat Diagn, 2021 Mar;41(4):505-516.
PMID: 33462877 DOI: 10.1002/pd.5903

OBJECTIVE: To investigate the performance of the machine learning (ML) model in predicting small-for-gestational-age (SGA) at birth, using second-trimester data.
METHODS: Retrospective data of 347 patients, consisting of maternal demographics and ultrasound parameters collected between the 20th and 25th gestational weeks, were studied. ML models were applied to different combinations of the parameters to predict SGA and severe SGA at birth (defined as 10th and third centile birth weight).
RESULTS: Using second-trimester measurements, ML models achieved an accuracy of 70% and 73% in predicting SGA and severe SGA whereas clinical guidelines had accuracies of 64% and 48%. Uterine PI (Ut PI) was found to be an important predictor, corroborating with existing literature, but surprisingly, so was nuchal fold thickness (NF). Logistic regression showed that Ut PI and NF were significant predictors and statistical comparisons showed that these parameters were significantly different in disease. Further, including NF was found to improve ML model performance, and vice versa.
CONCLUSION: ML could potentially improve the prediction of SGA at birth from second-trimester measurements, and demonstrated reduced NF to be an important predictor. Early prediction of SGA allows closer clinical monitoring, which provides an opportunity to discover any underlying diseases associated with SGA.
Fulltext The accuracy of international and national fetal growth charts in detecting small-for-gestational-age infants using the Lambda-Mu-Sigma method

Saw SN, Lim MC, Liew CN, Ahmad Kamar A, Sulaiman S, Saaid R, et al.

Front Surg, 2023;10:1123948.
PMID: 37114151 DOI: 10.3389/fsurg.2023.1123948

OBJECTIVE: To construct a national fetal growth chart using retrospective data and compared its diagnostic accuracy in predicting SGA at birth with existing international growth charts.
METHOD: This is a retrospective study where datasets from May 2011 to Apr 2020 were extracted to construct the fetal growth chart using the Lambda-Mu-Sigma method. SGA is defined as birth weight <10th centile. The local growth chart's diagnostic accuracy in detecting SGA at birth was evaluated using datasets from May 2020 to Apr 2021 and was compared with the WHO, Hadlock, and INTERGROWTH-21st charts. Balanced accuracy, sensitivity, and specificity were reported.
RESULTS: A total of 68,897 scans were collected and five biometric growth charts were constructed. Our national growth chart achieved an accuracy of 69% and a sensitivity of 42% in identifying SGA at birth. The WHO chart showed similar diagnostic performance as our national growth chart, followed by the Hadlock (67% accuracy and 38% sensitivity) and INTERGROWTH-21st (57% accuracy and 19% sensitivity). The specificities for all charts were 95-96%. All growth charts showed higher accuracy in the third trimester, with an improvement of 8-16%, as compared to that in the second trimester.
CONCLUSION: Using the Hadlock and INTERGROWTH-21st chart in the Malaysian population may results in misdiagnose of SGA. Our population local chart has slightly higher accuracy in predicting preterm SGA in the second trimester which can enable earlier intervention for babies who are detected as SGA. All growth charts' diagnostic accuracies were poor in the second trimester, suggesting the need of improvising alternative techniques for early detection of SGA to improve fetus outcomes.
Assessing the Accuracy of Fitness Smartwatch Data for Cardiovascular and Physical Activity Monitoring: A Validation Study in Digital Health

Mohamad Matrol AA, Koh M, Tan WPE, Ong HC, Ramanaidu LP, Saw SN

J Vis Exp, 2025 Feb 21.
PMID: 40063512 DOI: 10.3791/67674

This study aims to validate the accuracy of low-cost fitness smartwatches by comparing their data with gold-standard measurements for cardiovascular and physical activity parameters. The study enrolled 50 subjects, 26 undergoing validation testing for heart rate, blood oxygen saturation (SpO2), and sleep data against polysomnography (PSG). Additionally, 24 subjects participated in the 3-Minute Walk Test (3MWT) and Stairs Climbing (SC), with step counts validated against manual video calculations. Results showed no significant difference between the device's measurements and gold standard values for shallow sleep, deep sleep, REM time, mean heart rate, minimum heart rate, and SpO2. However, the device significantly underestimated manually counted steps (p = 0.009 (3MWT); p = 0.012 (SC)), total sleep duration (p = 0.004), and wake time (p = 8.94 × 10-8) while overestimating maximum heart rate (p = 0.011). These findings highlight the importance of accurate validation and interpretation of wearable device data in clinical contexts. Given these limitations, excluding the device's readings in future analyses is recommended to maintain data reliability and research integrity. This study underscores the need for ongoing validation and improvement of wearable technology to ensure its reliability and effectiveness in healthcare.
Enhancing Small-for-Gestational-Age Prediction: Multi-Country Validation of Nuchal Thickness, Estimated Fetal Weight, and Machine Learning Models

Deng J, Naresh Sethi NSA, Ahmad Kamar A, Saaid R, Loo CK, Mattar CNZ, et al.

Prenat Diagn, 2025 Mar;45(3):374-386.
PMID: 39817730 DOI: 10.1002/pd.6748

OBJECTIVE: The first objective is to develop a nuchal thickness reference chart. The second objective is to compare rule-based algorithms and machine learning models in predicting small-for-gestational-age infants.
METHOD: This retrospective study involved singleton pregnancies at University Malaya Medical Centre, Malaysia, developed a nuchal thickness chart and evaluated its predictive value for small-for-gestational-age using Malaysian and Singapore cohorts. Predictive performance using conjunctive (AND)/disjunctive (OR) rule-based algorithms was assessed. Seven machine learning models were trained on Malaysia data and evaluated on both Malaysia and Singapore cohorts.
RESULTS: 5519 samples were collected from the University Malaya Medical Centre. Small-for-gestational-age infants exhibit significantly lower nuchal thickness (small-for-gestational-age: 4.57 [1.04] mm, appropriate-for-gestational-age: 4.86 [1.06] mm, p
Fulltext Model learning analysis of 3D optoacoustic mesoscopy images for the classification of atopic dermatitis

Park S, Saw SN, Li X, Paknezhad M, Coppola D, Dinish US, et al.

Biomed Opt Express, 2021 Jun 01;12(6):3671-3683.
PMID: 34221687 DOI: 10.1364/BOE.415105

Atopic dermatitis (AD) is a skin inflammatory disease affecting 10% of the population worldwide. Raster-scanning optoacoustic mesoscopy (RSOM) has recently shown promise in dermatological imaging. We conducted a comprehensive analysis using three machine-learning models, random forest (RF), support vector machine (SVM), and convolutional neural network (CNN) for classifying healthy versus AD conditions, and sub-classifying different AD severities using RSOM images and clinical information. CNN model successfully differentiates healthy from AD patients with 97% accuracy. With limited data, RF achieved 65% accuracy in sub-classifying AD patients into mild versus moderate-severe cases. Identification of disease severities is vital in managing AD treatment.
Fulltext A deep learning-based pipeline for analyzing the influences of interfacial mechanochemical microenvironments on spheroid invasion using differential interference contrast microscopic images

Ngo TKN, Yang SJ, Mao BH, Nguyen TKM, Ng QD, Kuo YL, et al.

Mater Today Bio, 2023 Dec;23:100820.
PMID: 37810748 DOI: 10.1016/j.mtbio.2023.100820

Metastasis is the leading cause of cancer-related deaths. During this process, cancer cells are likely to navigate discrete tissue-tissue interfaces, enabling them to infiltrate and spread throughout the body. Three-dimensional (3D) spheroid modeling is receiving more attention due to its strengths in studying the invasive behavior of metastatic cancer cells. While microscopy is a conventional approach for investigating 3D invasion, post-invasion image analysis, which is a time-consuming process, remains a significant challenge for researchers. In this study, we presented an image processing pipeline that utilized a deep learning (DL) solution, with an encoder-decoder architecture, to assess and characterize the invasion dynamics of tumor spheroids. The developed models, equipped with feature extraction and measurement capabilities, could be successfully utilized for the automated segmentation of the invasive protrusions as well as the core region of spheroids situated within interfacial microenvironments with distinct mechanochemical factors. Our findings suggest that a combination of the spheroid culture and DL-based image analysis enable identification of time-lapse migratory patterns for tumor spheroids above matrix-substrate interfaces, thus paving the foundation for delineating the mechanism of local invasion during cancer metastasis.