MAMelkamu Hunegnaw Asmare

Melkamu Hunegnaw Asmare

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Melkamu Hunegnaw Asmare has a total of 31 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Deep Learning-Based Glaucoma Screening Using Smartphone-Captured Low-Quality Fundus Videos

    by Solomon Gebru Abay, Melkamu Hunegnaw Asmare, Julie Jacob, Lucca Geurts

    This study investigated the feasibility of deep learning–based glaucoma screening using fundus videos acquired with an ultra-low-cost, smartphone-based imaging system designed for resource-limited settings. Fundus videos were collected during routine clinical examinations using the glaucoma screening-on-phone (GSoP) adapter. A red-channel intensity-based strategy was employed to select diagnostically relevant frames from videos. Optic disc (OD) localizationwas performed using a fine-tunedYOLOv8n object detector, followed by OD-centered cropping and preprocessing. Glaucoma classification was conducted using MobileNetV3-Small model, fine-tuned via five-fold stratified group cross-validation. Performancewas evaluated using accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC), with additional testing on independent dataset. Comparative and ablation analyses assessed architectural choices and OD localization impact. Despite GSoP’s limited optical quality, the model achieved strong performance. On the independent test set, the single-frame MobileNetV3-Small reached an AUC of 0.91, with 93% sensitivity and 88% specificity. This semi-automated approach outperformed a fully automated multi-frame strategy (0.90 vs. 0.81 accuracy). Ablation analysis demonstrated that OD localization improved sensitivity and specificity by 13.3% and 9.5%, respectively, on MobileNetV3-Small model. The lightweight model performed comparably to deeper networks with significantly lower computational complexity. These findings indicate that clinically meaningful glaucoma-related features can be extracted from low-quality fundus images acquired with ultra-low-cost smartphone hardware, with semi-manual single-frame selection currently proving more effective than fully automated multi-frame-based prediction. This work serves as an encouraging proof-of-concept for combining affordable imaging systems with efficient deep learning pipelines to achieve scalable glaucoma screening in underserved settings.

  • Open Access

    Rheumatic Heart Disease Detection in Asymptomatic Schoolchildren using ECG and PCG

    by Amsalu Tomas Chuma, Chunzhuo Wang, Jens-Uwe Voigt, Desalew Mekonnen Kassie, Melkamu Hunegnaw Asmare, Bart Vanrumste

    Background: Rheumatic heart disease (RHD) remains a major public health concern across low- and middle-income countries in the Global South. Early detection through community-based screening of asymptomatic individuals has been identified as a critical strategy for reducing the disease burden. Despite this, the absence of accessible, automated population screening tools continues to impede implementation at scale. This study investigates the screening potential of integrating electrocardiography (ECG) and phonocardiography (PCG) for the early detection of RHD in asymptomatic schoolchildren. Methods: The dataset was obtained as part of an ambulatory screening initiative conducted across multiple school sites in rural areas of Ethiopia. It comprised ECG and PCG recordings from 611 asymptomatic schoolchildren aged 10 to 20 years. A comprehensive set of time–frequency, visibility graph and non-linear features were extracted from both signal modalities. These features were subsequently evaluated using machine learning models to assess their utility in the automated screening of early RHD. Results: The best model achieved an average 10-folds cross-validation scores on sensitivity, positive-predictive-value and F1-score of 59.6%, 63.6% and 60.8%, respectively for multimodal ECG and PCG signals. Whereas separate evaluation of ECG showed an F1-score of 61.1% and PCG achieved 23.5%. Key features included the T-wave, the area under the QRS complex, and entropy measures derived from beat visibility graphs in the ECG. In addition, visibility graph features from multi-band S1 and S2 heart sound segments, along with MFCC coefficients from the PCG, were also relevant. However, PCG alone performed poorly and did not show improved results over the ECG features. Conclusion: Although auscultation is key clinical diagnosis tool in symptomatic RHD, combined PCG with ECG features does not enhance asymptomatic RHD detection using the ECG modality alone.

  • Open Access

    Screening for Rheumatic Heart Disease in Asymptomatic Children using Machine Learning from Electrocardiograms

    by Amsalu Tomas Chuma, Chunzhuo Wang, Melkamu Hunegnaw Asmare, Carolina Varon, Jens-Uwe Voigt, Desalew Mekonnen Kassie, Liesl Zühlke, Bart Vanrumste

    Early detection of Rheumatic Heart Disease (RHD) is essential in reducing its associated mortality and late complications. In resource-limited settings, automated detection using low-cost electrocardiogram (ECG) sensors can enhance prevention efforts. However, its effectiveness as a potential RHD screening tool in at-risk populations remains unexplored. This study aimed to investigate the utility of machine learning for classifying RHD in a cohort screened for RHD using low-cost ECG devices. The ECGs were collected from 611 at-risk schoolchildren using KardiaMobile, where 47 were confirmed RHD and 564 were healthy. First, the ECG fiducial points were annotated using a publicly available prominence-based delineator. Then, temporal, frequency, wavelet, and visibility graph-based features were extracted from six-leads and fed to the XGBoost classifier. A 10-fold cross-validation was used at different prediction score thresholds to obtain target sensitivity (Se) for screening RHD. Single-lead evaluation on Lead-II showed an F1-score of 60.9%, a Se of 59.6% and a positive-predictive-value (PPV) of 62.2%. However, using multiple leads improved the results, with an F1-score of 62.8%, a Se of 59.6% and a PPV of 66.7%. The best model performance was achieved by adjusting the threshold to 0.6 with Se and PPV of 66% and 51%, respectively. Error analysis revealed that T-wave and STT changes, as well as non-rheumatic mitral valve cases were among the false positive cases. Machine learning can enhance early detection by leveraging relevant ECG features and adjustable target sensitivity based on screening priorities and resource capacity. Measurements can be obtained without chest contact, using only the fingers and knees, thereby enabling use by non-clinical staff. This approach provides a scalable and cost-effective solution for RHD screening in high-prevalence regions.

  • Open Access

    Multicenter Chest Radiographic Findings from Ethiopia with Radiologist-Generated Annotations

    by Messay Gebrekidan, Samuel S. Hailu, Michael Teklehaimanot Abera, Melkamu Hunegnaw Asmare, Michael A. Negussie

    Background Chest radiography is widely used in Ethiopia for the evaluation of respiratory and cardiac disease. However, chest X-ray datasets used to develop and benchmark artificial intelligence (AI) systems are predominantly derived from high-income settings, which may limit generalizability and obscure region-specific radiographic patterns. Purpose To describe multicenter patterns of chest X-ray imaging findings in Ethiopia using the Afro-Chest X-ray cohort and to summarize the radiologist-led workflow used to generate high-quality localization labels. Materials and Methods This retrospective multicenter study included deidentified chest X-rays acquired at 10 Ethiopian institutions from December 2022 through July 2025. After quality filtering, 55,409 chest X-rays were retained. Radiology reports were standardized using the Afro-Chest X-ray reporting template (Table 5 ). A stratified subset of 11,880 chest X-rays was manually annotated by 11 radiologists using bounding boxes for 19 thoracic findings (Table 3) with recorded confidence levels. Finding patterns was summarized descriptively using counts and proportions at the exam and finding-instance levels. Results Among 55,409 chest X-rays, 31,939 were linked to radiology reports from 48,962 patients (male, 18,324 [37.4%]; female, 30,387 [62.1%]). In the annotated subset (11,880 chest X-rays), 7,003 (58.9%) were abnormal and contained 22,531 labeled finding instances (mean, 3.22 instances per abnormal chest X-ray). The most frequent findings were reticulonodular or ground-glass opacities (n = 6,896, 30.6%), pleural effusion (n = 2,685, 11.9%), cardiomegaly or chamber enlargement (n = 2,161, 9.6%), fibrosis or fibrobronchiectatic change (n = 1,869, 8.3%), and consolidation with cavitary lesions (n = 1,835, 8.1%). Conclusion In this multicenter Ethiopian cohort, annotated abnormalities were dominated by parenchymal opacities, pleural effusion, and cardiomegaly. Afro-Chest X-ray provides a radiologist-verified reference for describing regional chest X-ray patterns and supporting the local validation and development of AI systems applicable to East and other sub-Saharan African clinical settings.

  • Open Access

    Prevalence of Early Rheumatic Heart Disease Among Asymptomatic Students in Underserved Communities in Ethiopia: Cross-Sectional Observational Study

    by Amsalu Tomas Chuma, Desalew Mekonnen Kassie, Jens-Uwe Voigt, Melkamu Hunegnaw Asmare, Ahmed Saeed Youssef, Carolina Varon, Rik Willems, Michelle Yates, Adane Petros Sikamo, Yidnekachew Asrat Birhan, et al.

    Background: Rheumatic heart disease (RHD) is a sequela of recurrent, untreated group A Streptococcus infections. RHD disproportionately affects children and young adults in the Global South. Intermittent mass screening of early RHD by using affordable tools in these disease-endemic regions is essential for effective prevention. Objective: This study examined multimodal physiological data for assessing the prevalence of early RHD in a cohort of asymptomatic, at-risk students in rural Ethiopia. Methods: A total of 584 asymptomatic children aged 10 to 20 years were randomly selected for screening and stratified into 2 groups (≤14 and >14 years). Electrocardiogram, phonocardiogram, and echocardiography screening were performed, with diagnoses based on the 2012 World Heart Federation criteria. Results: After excluding 1.4% (8/584) of the children, who had nonrheumatic findings, 576 participants were analyzed, including 334 (58.0%) female and 242 (42.0%) male children. The mean age was 16.1 (SD 2.4) years (95% CI 15.9-16.3). Echocardiographic screening identified 19 cases of RHD (n=10, 52.6% borderline and n=9, 47.4% definite). Female children accounted for 68.4% (13/19) of cases, and the association between female sex and RHD was not statistically significant (odds ratio 1.59, 95% CI 0.60-4.25; P=.35). The estimated prevalence of RHD was 32.5 per 1000 population (95% CI 18.1-46.9; SE 7.4; 19/576, 3.3%), which was significantly higher than the most recent multicenter prevalence estimate of 19 per 1000 population (95% CI 13.9-23.4; odds ratio 2.12; P=.03). Mitral regurgitation was the predominant lesion (16/19, 84.2%), followed by mitral stenosis (2/19, 10.5%) and aortic regurgitation (1/19, 5.3%). Phonocardiogram analysis showed mitral regurgitation (10/19, 52.6%), mitral stenosis (2/19, 10.5%), and subclinical findings in the rest of patients with RHD. Prolonged PR intervals were observed in 10.5% (2/19) of the RHD-positive participants. Conclusions: This study confirms persistent high prevalence of asymptomatic RHD among students in rural regions of Ethiopia. Although there was a female predominance in RHD incidence, the difference between the sexes was not statistically significant.


  • Open Access

    Performance Assessment of ECG Delineators on Single-Lead Wearable Ambulatory Data

    by Amsalu Tomas Chuma, Ahmed Saeed Youssef, Melkamu Hunegnaw Asmare, Chunzhuo Wang, Desalew Mekonnen Kassie, Jens-Uwe Voigt, Bart Vanrumste

    Reliable interpretation of electrocardiograms (ECGs) requires precise identification of P, QRS, and T (PQRST) wave boundaries. However, it remains challenging due to noise, signal quality variability, and inherent morphological diversity particularly in recordings from children. This study systematically compares the performance of leading deep neural networks (DNN) and heuristic-based delineation algorithms on ambulatory single-lead ECG signals focusing on temporal accuracy. Experiments were conducted using the publicly available LUDB dataset and a private validation dataset comprising 21,759 annotated single-lead wave segments from 611 children recorded using KardiaMobile ECG sensor. DNN were first trained on the LUDB dataset and subsequently tested on the validation dataset. The delineation performance was assessed using Sensitivity (Se) and positive-predictive-value (P + ) metrics. The best-performing heuristic based and DNN models reached Se and P + of (98.9% vs 97.9%) for P, (99.8% vs 99.2%) for QRS, and (98.7% vs 95.9%) for T wave fiducials, respectively. The lowest standard-deviation (in ms) of wave onset/offset delineation was achieved by attention based 1DU-Net model; ±16.6/±16.3 for P-wave, ±14.0/±16.3 for QRS, and ±26.3/±18.8 for T-wave, respectively. The findings indicate that optimized heuristic models can perform comparably to complex DNN, highlighting their efficiency and suitability for real-time ECG delineation in digital health monitoring applications.


  • Open Access

    Feasibility and early clinical yield of handheld AI-assisted obstetric point-of-care ultrasound in routine antenatal care: a pilot study in Addis Ababa, Ethiopia

    by Crystal Richardson, Messay Gebrekidan, Melkamu Hunegnaw Asmare, Daniel Cherkos Teka, Saba Habtu Hagos, Eilham Elias Jobir, Abenezer Shiferaw, Tsega Woldegiorgis Godebo, Akrem Kedir Shemsu

    Background: The World Health Organization recommends at least one ultrasound examination before 24 weeks of gestation to improve gestational-age assessment and detection of multiple pregnancy and fetal anomalies. However, access to obstetric ultrasound at the primary-care level remains limited in many low-resource settings, including Ethiopia. This study evaluated the feasibility and early clinical yield of integrating AI-enabled handheld obstetric point-of-care ultrasound (POCUS) into routine antenatal care in four government health centers in Addis Ababa. Methods: We conducted a 4-week prospective pilot across four urban health centers (Abuare, Signal-Woreda 7, Kebena, and Janmeda). Pregnant women attending routine antenatal care (ANC) underwent focused obstetric POCUS using Butterfly handheld devices. Trained frontline providers, mainly junior GPs and midwives, performed scans using a standardized acquisition form, while clinical data were entered into REDCap and linked to archived image records on the Butterfly cloud platform. Primary outcomes were the feasibility of workflow integration, completeness of focused scan variables, the spectrum of focused obstetric findings, and referrals generated. Results: A total of 101 women were scanned. Mean maternal age was 28.3 years (SD 4.4); median gravidity was 2 (IQR 1–3), and median parity was 1 (IQR 0–2). Ultrasound-estimated gestational age was recorded in 100/101 women (median 30 weeks; IQR 16.8–33.0). Fetal cardiac activity was present in 92/99 (92.9%), absent in 4/99 (4.0%), and indeterminate in 3/99 (3.0%). Fetal presentation was cephalic in 65/101 (64.4%), breech in 5/101 (5.0%), transverse in 17/101 (16.8%), and indeterminate in 14/101 (13.9%). Placental location was low-lying in 1/100 (1.0%), and qualitative amniotic fluid was low in 1/100 (1.0%). Seven of 99 women (7.1%) were referred to a higher-level facility. A binary normal-scan indicator classified 93/101 scans (92.1%) as normal. Conclusions: Handheld focused obstetric POCUS was operationally feasible within routine urban ANC and identified a clinically relevant minority of women requiring referral or further review. Larger studies with standardized expert over-read, referral follow-up, and explicit image-quality metrics are needed before stronger claims about referral optimization or broader clinical impact can be made.


  • Open Access

    Electrocardiographic Digital Biomarkers in Asymptomatic Schoolchildren with Rheumatic Heart Disease

    by Amsalu Tomas Chuma, Jens-Uwe Voigt, Ahmed Sayeed Youssef, Melkamu Hunegnaw Asmare, Chunzhuo Wang, Carolina Varon, Desalew Mekonnen Kassie, Rik Willems, Bart Vanrumste

    Subclinical rheumatic valvular disease is a significant yet underdiagnosed contributor to the global rheumatic heart disease (RHD) burden. Early detection through population screening is essential to prevent its progression to severe RHD. Rhythm changes and prolongations of PR and QTc intervals in the ECG are described in the advanced RHD cases. However, these parameters were not yet studied in asymptomatic RHD. We aimed to investigate the potential of ECG biomarkers for screening RHD in asymptomatic schoolchildren. ECG tracings from 611 schoolchildren aged 10 to 20 years were selected from a cohort screened for RHD in four schools in an RHD-endemic region. Confirmatory diagnoses were based on echocardiographic findings, where 564 (F=326, M=238) were healthy, and 47 (F=28, M=19) were positive for RHD (24 borderline RHD and 23 definite RHD). Independent, blinded reviewers manually annotated the ECGs and PR interval (PR), P-wave dispersion (PWd), and the ratio between the P-wave duration and PR interval (Pw/PR) were analyzed. The mean age of the study cohort at diagnosis was 16.1 ± 2.5 years, and 58% of the participants were females. Atrial fibrillation was seen in 8% (n=4), and prolonged PR in 2% (n=1) of RHD-positive cases. The mean ± std for normals vs RHD is (PR, 138±19 vs 150±19), (Pw/PR, 0.75±0.06 vs 0.71±0.07), and (PWd, 49±14 vs 56±17). The PR (p<0.001), Pw/PR (p<0.001), and PWd (p=0.008) showed a significant difference between healthy and RHD-positive subjects. The PR was increased consistently with severity across age groups above and below 16 years. The PR, PWd, and Pw/PR can serve as non-invasive biomarkers for the screening of RHD in at-risk schoolchildren. Monitoring alterations in these markers at an early stage of RHD is crucial for enabling prompt management and follow-up. It is thus evident that ECG can support an intermittent ambulatory RHD screening in resource-limited settings.

  • Open Access

    Prevalence of Early Rheumatic Heart Disease Among Asymptomatic Students in Underserved Communities in Ethiopia: Cross-Sectional Observational Study (Preprint)

    by Amsalu Tomas Chuma, Desalew Mekonnen Kassie, Jens-Uwe Voigt, Melkamu Hunegnaw Asmare, Ahmed Saeed Youssef, Carolina Varon, Rik Willems, Michelle Yates, Adane Petros Sikamo, Yidnekachew Asrat Birhan, et al.

    BACKGROUND Rheumatic heart disease (RHD) is a sequela of recurrent, untreated group A Streptococcus infections. RHD disproportionately affects children and young adults in the Global South. Intermittent mass screening of early RHD by using affordable tools in these disease-endemic regions is essential for effective prevention. OBJECTIVE This study examined multimodal physiological data for assessing the prevalence of early RHD in a cohort of asymptomatic, at-risk students in rural Ethiopia. METHODS A total of 584 asymptomatic children aged 10 to 20 years were randomly selected for screening and stratified into 2 groups (≤14 and >14 years). Electrocardiogram, phonocardiogram, and echocardiography screening were performed, with diagnoses based on the 2012 World Heart Federation criteria. RESULTS After excluding 1.4% (8/584) of the children, who had nonrheumatic findings, 576 participants were analyzed, including 334 (58.0%) female and 242 (42.0%) male children. The mean age was 16.1 (SD 2.4) years (95% CI 15.9-16.3). Echocardiographic screening identified 19 cases of RHD (n=10, 52.6% borderline and n=9, 47.4% definite). Female children accounted for 68.4% (13/19) of cases, and the association between female sex and RHD was not statistically significant (odds ratio 1.59, 95% CI 0.60-4.25; P=.35). The estimated prevalence of RHD was 32.5 per 1000 population (95% CI 18.1-46.9; SE 7.4; 19/576, 3.3%), which was significantly higher than the most recent multicenter prevalence estimate of 19 per 1000 population (95% CI 13.9-23.4; odds ratio 2.12; P=.03). Mitral regurgitation was the predominant lesion (16/19, 84.2%), followed by mitral stenosis (2/19, 10.5%) and aortic regurgitation (1/19, 5.3%). Phonocardiogram analysis showed mitral regurgitation (10/19, 52.6%), mitral stenosis (2/19, 10.5%), and subclinical findings in the rest of patients with RHD. Prolonged PR intervals were observed in 10.5% (2/19) of the RHD-positive participants. CONCLUSIONS This study confirms persistent high prevalence of asymptomatic RHD among students in rural regions of Ethiopia. Although there was a female predominance in RHD incidence, the difference between the sexes was not statistically significant.