Dual concatenated transfer learning with attention fusion: An ensemble-enhanced approach for skin lesion classification - Takeaways - MDSpire

Dual concatenated transfer learning with attention fusion: An ensemble-enhanced approach for skin lesion classification

  • By

  • Probal Bhowmick

  • Julia Rahman

  • Anwar Hossain Efat

  • Tasfi Fairoz Nidhi

  • Dipanjan Karmaker Amit

  • June 8, 2026

  • 0 min

Share

  • 1

    Skin lesions are critical indicators of dermatological disorders, with melanoma being the most lethal skin cancer type.

  • 2

    AI, particularly ML and DL, can enhance self-screening for skin cancer, addressing the challenges of high consultation costs.

  • 3

    The study introduced a hybrid multi-branch ensemble framework combining DenseNet and ResNet variants to improve classification performance.

  • 4

    Attention mechanisms were integrated to enhance feature extraction and reduce background noise in skin lesion images.

  • 5

    A comprehensive evaluation using multiple metrics was conducted to ensure model validity, particularly for imbalanced datasets.

Original Source(s)

Dual concatenated transfer learning with attention fusion: An ensemble-enhanced approach for skin lesion classification

  • by Probal Bhowmick, Julia Rahman, Anwar Hossain Efat, Tasfi Fairoz Nidhi, Dipanjan Karmaker Amit

  • June 8, 2026

Related Content

Tumor-localized CD40 agonism with MP0317, a FAP x CD40 DARPin, reprograms the tumor microenvironment in patients with advanced solid tumors: an open-label, nonrandomized, dose-escalation phase 1 study

High-frequency ultrasound combined with superb microvascular imaging for differential diagnosis of basal cell carcinoma and seborrheic keratosis: a retrospective analysis with nomogram implementation

Author Correction: The efficacy of chemotherapy is limited by intratumoral senescent cells expressing PD-L2