Ayush Somani

UiT The Arctic University of Norway · Department of Computer Science

Accurate classification of microscopy images is critical for the analysis of biological samples. The availability of large-scale labeled datasets has contributed to recent progress in training large, deep classification models in the medical imaging domain, but methods that cater to a variety of microscopy modalities across a range of biological sa...

Scanning acoustic microscopy (SAM) is a cutting-edge label-free imaging technique that allows viewing of both surface and internal structures in a variety of samples, including industrial and biological. Several factors influence the acoustic image resolution, including the signal-to-noise ratio, scanning step size, and the transducer frequency. Ou...

Artificial networks of neurons and algorithms mimic human neural systems intertwined and stacked. Most can’t comprehend a human mind, yet they work flawlessly until they don’t. With millions of billions of parameter optimizations in hundreds of layers over a short time, the question is not ‘if’ something will go wrong, but ‘when’.

When it comes to interpretability, ML models, particularly DL models, are frequently regarded as a black box due to their complexity and lack of transparency in approach. It is fairly simple to train a network to be specific. A DL model learns to classify an object, recognize text, or generate digital images. It efficiently encapsulates feature lea...

Artificial Intelligence (AI) and modern computing captivate a large and growing number of people. It’s fascinating to see how they progressed from a mere impression of mimicking human-like behavior to surpassing human-level performance that fits in one’s pocket.

Previous chapters have discussed the learning, performance and explainability of NNs based on crisp inputs, weights, parameters, training samples and other information pieces. It was demonstrated how a deep network propagates information into its layers.

The chapter commences with a prevalent phrase in the modern era “AI will take over the world!” (Crawford and Calo 2016). There are two major interpretations of the phrase. The first is to acknowledge AI as a developing technology with the ability to cater to larger organizations, global automation, and streamline inefficient procedures. The other i...

Mitochondria play a crucial role in cellular metabolism. This paper presents a novel method to visualize mitochondria in living cells without the use of fluorescent markers. We propose a physics-guided deep learning approach for obtaining virtually labeled micrographs of mitochondria from bright-field images. We integrate a microscope’s point sprea...

We often locate ourselves in a trade-off situation between what is predicted and understanding why the predictive modeling made such a prediction. This high-risk medical segmentation task is no different where we try to interpret how well has the model learned from the image features irrespective of its accuracy. We propose image-specific fine-tuni...

Examining specific sub-cellular structures while minimizing cell perturbation is important in the life sciences. Fluorescence labeling and imaging is widely used for introducing specificity despite its perturbative and photo-toxic nature. With the advancement of deep learning, digital staining routines for label-free analysis have emerged as a repl...