Bayesian Convolutional Neural Network-based Models for Diagnosis of Blood Cancer

ABSTRACT

Deep learning methods allow computational models involving multiple processing layers to discover intricate structures in data sets. Classifying an image is one such problem where these methods are found to be very useful. Although different approaches have been proposed in the literature, this paper illustrates a successful implementation of the Bayesian Convolution Neural Networks (BCNN)-based classification procedure to classify microscopic images of blood samples (lymphocyte cells) without involving manual feature extractions. The data set contains 260 microscopic images of cancerous and noncancerous lymphocyte cells. We experiment with different network structures and obtain the model that returns the lowest error rate in classifying the images. Our developed models not only produce high accuracy in classifying cancerous and noncancerous lymphocyte cells but also provide useful information regarding uncertainty in predictions.

Acknowledgments

Farrukh Javed acknowledges financial support from the internal research grants at Örebro University.

Data Availability Statement

The data that support the findings of this study are openly available in [repository: ALL_IDB] and can be accessed at https://homes.di.unimi.it/scotti/all/.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplemental data

Supplemental data for this article can be accessed on the publisher’s website.

Notes

1. In McLachlan et al. (2020), the authors summarized the Bayesian approaches used in healthcare in producing meaningful and accurate decision-support systems.

2. For more technical details, we suggest readers to please refer to the references therein.

3. See Goodfellow, Bengio, and Courville (2016) for details.

4. Several networks are tested during the process but based on the performances, only six networks are kept for the rest of the analysis.

5. We report the highest reported overall accuracy here. For detailed results, we refer the readers to Rawat et al. (2017); Putzu, Caocci, and Di Ruberto (2014).

6. Note that all other models from respective network structures showed almost a similar performance with slight deviations.