Deep optical neural network by living tumour brain cells

D. Pierangeli,V. Palmieri,G. Marcucci,C. Moriconi,G. Perini,M. De Spirito,M. Papi,C. Conti

DOI: https://doi.org/10.48550/arXiv.1812.09311

2018-12-09

Abstract:The new era of artificial intelligence demands large-scale ultrafast hardware for machine learning. Optical artificial neural networks process classical and quantum information at the speed of light, and are compatible with silicon technology, but lack scalability and need expensive manufacturing of many computational layers. New paradigms, as reservoir computing and the extreme learning machine, suggest that disordered and biological materials may realize artificial neural networks with thousands of computational nodes trained only at the input and at the readout. Here we employ biological complex systems, i.e., living three-dimensional tumour brain models, and demonstrate a random neural network (RNN) trained to detect tumour morphodynamics via image transmission. The RNN, with the tumour spheroid as a three-dimensional deep computational reservoir, performs programmed optical functions and detects cancer morphodynamics from laser-induced hyperthermia inaccessible by optical imaging. Moreover, the RNN quantifies the effect of chemotherapy inhibiting tumour growth. We realize a non-invasive smart probe for cytotoxicity assay, which is at least one order of magnitude more sensitive with respect to conventional imaging. Our random and hybrid photonic/living system is a novel artificial machine for computing and for the real-time investigation of tumour dynamics.

Medical Physics,Disordered Systems and Neural Networks,Adaptation and Self-Organizing Systems,Biological Physics,Optics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the need to develop large - scale ultra - fast hardware to support machine learning in the new era of artificial intelligence. Specifically, the paper explores how to use biological materials, especially the living three - dimensional tumor brain cell model, to construct a recurrent neural network (RNN). This network can detect the morphodynamics of tumors and detect cancer morphodynamic changes that cannot be obtained by optical imaging during laser - induced hyperthermia. In addition, this RNN can also quantify the effect of chemotherapy on inhibiting tumor growth, achieving a non - invasive intelligent probe for cytotoxicity determination, with a sensitivity at least one order of magnitude higher than that of traditional imaging techniques. This research provides a new bio/photonic hybrid computing system for real - time investigation of tumor dynamics.

Deep optical neural network by living tumour brain cells

Optical Implementation of Neural Learning Algorithms Based on Cross-gain Modulation in a Semiconductor Optical Amplifier

Dense 4D nanoscale reconstruction of living brain tissue

Deep learning with coherent VCSEL neural networks

Photonic extreme learning machine by free-space optical propagation

Ti$_3$C$_2$T$_x$ MXene Enabled All-Optical Nonlinear Activation Function for On-Chip Photonic Deep Neural Networks

All-optical spiking neurosynaptic networks with self-learning capabilities

Retinomorphic Machine Vision in a Network Laser

Optical neuromorphic processing at Tera-OP/s speeds based on Kerr soliton crystal microcombs

Resolution-enhanced multi-core fiber imaging learned on a digital twin for cancer diagnosis

Large-scale photonic computing with nonlinear disordered media

Multiwavelength Neuromorphic Silicon Photonics

High-fidelity multimode fibre-based endoscopy for deep-brain in vivo imaging

Deep intravital brain tumor imaging enabled by tailored three-photon microscopy and analysis

Photonic neural networks based on integrated silicon microresonators

Intra-operative Brain Tumor Detection with Deep Learning-Optimized Hyperspectral Imaging

Neuromorphic Silicon Photonics and Hardware-Aware Deep Learning for High-Speed Inference

Deep Learning for Design and Retrieval of Nano-photonic Structures

A complete, parallel and autonomous photonic neural network in a semiconductor multimode laser

Single photonic perceptron based on a soliton crystal Kerr microcomb for high-speed, scalable, optical neural networks