MedElevator Войти в исследования
← Поиск публикаций

DOI 10.1038/s41598-026-60525-7

Медицинская библиотека · Scientific reports · 2026

Superfast prediction of brain tumours through adaptive incremental pre-training and re-training through transfer learning using state-of-the-art ResNet18 architectural model.

Pinnaboina Sivani, Kambhampati Venkata Sowmya, Jammalamadaka Kodanda Rama Sastry, Jammalamadaka Sasi Bhanu

СтатьяПеревод в очереди

Аннотация

Brain tumour classification from Magnetic Resonance Imaging (MRI) requires both high diagnostic accuracy and computationally efficient model adaptation. Although deep transfer learning has improved automated tumour recognition, most existing approaches remain limited by dependence on generic ImageNet-pretrained weights, prolonged fine-tuning time, and weak institution-specific adaptability under small local medical datasets. To address these limitations, this paper proposes an Adaptive Incremental Domain Pretraining and Frozen-Weight Local Rapid Adaptation framework built on the ResNet18 backbone. Unlike conventional one-step transfer learning, the proposed method recursively preserves and updates tumour-specialised parameter states across multiple same-domain MRI repositories, progressively constructing a domain-adapted diagnostic backbone. The final inherited model is then subjected to frozen-weight local rapid adaptation for institution-specific MRI customisation. Experimental evaluation shows that direct ImageNet-based ResNet18 transfer learning achieves 93.27.

Для цитирования: Pinnaboina Sivani, Kambhampati Venkata Sowmya, Jammalamadaka Kodanda Rama Sastry, Jammalamadaka Sasi Bhanu. Superfast prediction of brain tumours through adaptive incremental pre-training and re-training through transfer learning using state-of-the-art ResNet18 architectural model.. Scientific reports. 2026. DOI: 10.1038/s41598-026-60525-7.

Предметные рубрики MeSH

Подготовка и перевод публикации
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…

История · было → стало

Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.

Тип публикации указан по данным источника. Уровень доказательности автоматически не присваивается.