Background: Chagas disease induces both focal and diffuse ECG abnormalities, such as bundle branch blocks and inferior Q waves among others. In resource-limited settings, where confirmatory testing is scarce, deep learning models can support triage. The PhysioNet Challenge 2025 emphasizes prioritization by evaluating models using true positive rate in the top 5% of predictions (TPR@5%). Objective: To develop a lead-aware deep learning model for detecting Chagas disease from ECGs and ranking high-risk patients for confirmatory testing. Methods: We developed LA-ECGNet, a convolutional neural network that uses depth-wise convolutions for lead-specific temporal feature extraction and pointwise convolutions for inter-lead fusion. The model was pretrained on a subset of the CODE-15% dataset (~300,000 ECGs with weak labels) using two self-supervised tasks: (i) reconstruction of randomly masked temporal segments within selected leads (masked segment reconstruction), and (ii) prediction of a held-out lead using the remaining 11 (lead prediction). Only a subset was used due to resource constraints during initial development. Fine-tuning was performed using binary cross-entropy (BCE) loss on a moderately imbalanced dataset with 1,631 serologically confirmed positives from SaMi-Trop and 6,524 presumed negatives from PTB-XL. Stratified 5-fold cross-validation ensured non-overlapping samples and label balance. Preprocessing included standardization to 10-second ECGs and per-lead normalization. Results: A partial implementation of LA-ECGNet trained on a small subset of the data achieved a challenge score of 0.84 and accuracy of 0.92 in local evaluation. However, due to time constraints, this model could not be submitted to the competition leaderboard. Conclusion: LA-ECGNet demonstrated strong potential in effectively modeling lead-specific ECG features with the aid of self-supervised pretraining. Future work will focus on full-cohort training, ranking-aware loss functions, calibration, explainability, and ensembling to improve generalizability and clinical impact in Chagas screening.