SZSyed Sajjad Imam Zaidi

Syed Sajjad Imam Zaidi

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

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Syed Sajjad Imam Zaidi has a total of 5 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Optimizing Transfer Learning for Melanoma Detection: A Fine-Tuned CNN Model with Robust Predictive Accuracy

    by Adnan Afroz, Muhammad Umar Khan, Umair Jilani, Syed Umair Naqvi, Syed Sajjad Imam Zaidi, Ali Yousuf Khan

    Early and precise identification is essential for improving patient prognosis because melanoma is still one of the most aggressive and deadly dermatological cancers. This paper suggests an improved deep learning framework that uses a refined Convolutional Neural Network (CNN) based on InceptionV3 to automatically classify skin lesions. The model tackles the intrinsic complexity of dermoscopic images by combining transfer learning with sophisticated data augmentation and tailored training hyperparameters. Over 100 training epochs, the method was assessed using a binary dataset of benign and malignant tumors. Metrics including accuracy, precision, sensitivity, specificity, and Area Under the Curve (AUC) were used to thoroughly evaluate performance. According to experimental data, the suggested model obtains a high classification accuracy of around 94%, with sensitivity and F1-scores of 93% respectively. Additionally, even in the presence of class imbalance, the model's resilience and discriminative capacity are confirmed by ROC and Precision-Recall assessments. When compared to baseline designs and modern state-of-the-art CNN models like ResNet50, VGG16, and EfficientNet, the proposed approach demonstrates competitive performance with improved sensitivity and specificity balance while maintaining reduced false positive and false negative predictions. By merging an enhanced InceptionV3 architecture with effective transfer learning, adaptive hyperparameter tuning, and enhanced augmentation strategies, the proposed framework provides a more balanced and dependable melanoma classification performance than conventional CNN-based methods. These results highlight the potential of the improved InceptionV3 model as a trustworthy computer-aided diagnostic (CAD) tool, giving dermatologists vital decision help in the early identification of melanoma.