A multimodal scanning electron microscopy-atomic force microscopy images for early cervical cancer detection using...

Area of Science:

• Biomedical imaging
• Nanotechnology
• Cancer diagnostics

Background:

• Early diagnosis of cervical cancer is crucial for effective treatment.
• Current diagnostic methods may have limitations in reliability.
• Combining imaging modalities offers potential for improved accuracy.

Purpose of the Study:

• To enhance early cervical cancer diagnosis by integrating scanning electron microscopy (SEM) and atomic force microscopy (AFM) data.
• To develop a multimodal approach overcoming the limitations of single imaging techniques.

Main Methods:

• Utilized 540 SEM and 540 AFM cell images.
• Applied discrete wavelet transform for feature extraction.
• Employed triangle, Jensen-Shannon, and Hellinger divergence classifiers.
• Incorporated exponential graph regulation for feature likelihoods and weighted classification.

Main Results:

• Triangle divergence showed superior performance among the classifiers.
• The weighted feature approach enhanced diagnostic precision.
• The combined SEM and AFM method accurately predicted cervical cancer likelihood.

Conclusions:

• The multimodal classification using triangle divergence significantly outperforms single-modality approaches.
• The integrated SEM and AFM method achieved 95.9% classification accuracy.
• This approach demonstrates high effectiveness for early cervical cancer detection.