MedBin: A lightweight End-to-End model-based method for medical waste management

Area of Science:

• Environmental Science
• Computer Science
• Healthcare Management

Background:

• Increasing volumes of medical waste necessitate advanced, cost-effective management solutions.
• Current medical waste management practices often lack automation and sophisticated categorization.
• The need for sustainable practices in healthcare is paramount.

Purpose of the Study:

• To introduce MedBin, a novel approach for automated medical waste sorting, reusing, and recycling.
• To develop and validate the MedBin-Net model for accurate medical waste detection and segmentation.
• To establish a comprehensive dataset for training and evaluating medical waste management systems.

Main Methods:

• Creation of the MedBin-Dataset with 2,119 images across 36 categories.
• Development of the lightweight MedBin-Net model for instance segmentation of medical waste.
• Implementation of a World Health Organization (WHO) Guideline-Based Classifier for waste categorization and disposal recommendations.

Main Results:

• The MedBin-Net model achieved high performance with an average precision of 0.91, recall of 0.97, and F1-score of 0.94.
• The model is computationally efficient, featuring 2.51 million parameters and an inference time of 0.60 ms.
• The WHO Guideline-Based Classifier accurately categorizes waste into 5 types with appropriate disposal methods.

Conclusions:

• The proposed MedBin system offers a promising, automated solution for sustainable medical waste management.
• The MedBin-Dataset and MedBin-Net model provide valuable resources for advancing research in this field.
• This approach supports adherence to international standards for medical waste disposal.