Deep Learning Glossary by NVIDIA®

Angiography uses contrast media injected into the blood vessels in combination with x-rays to visualize the inside or lumen of blood vessels, particularly the arteries, veins and the heart chambers. Example of clinical applications: diagnose obstructive vascular disease, bleeding vessels, aneurisms.

Caffe, Caffe2
Caffe was originally developed by the Berkeley Vision and Learning Center (BVLC) and by community contributors. It is a deep learning framework made with expression, speed, and modularity in mind.

(Microsoft) Cognitive Toolkit, CNTK
Microsoft Cognitive Toolkit, previously known as CNTK and sometimes styled as The Microsoft Cognitive Toolkit, is a deep learning framework developed by Microsoft Research.

Computed tomography (CT)
Computed tomography (CT) uses x-ray photons (with or without a contrast media) for image production with digital reconstruction to create a 2 or 3-dimensional image. Analog data captured by the scanner is digitally converted by various algorithms into reconstructed images which represent a cross-sectional "slice" through the patient. They provide more detailed information than x-rays. Example of clinical applications: diagnose muscle, bone disorders (tumors, fractures); pinpoint location of tumor, infection or clot, guide procedures (surgery, biopsy); detect internal injuries and bleeding.

Conventional radiography: use of x-rays, a form of electromagnetic radiation produced by an x-ray tube, to visualize the internal structures of a patient. The variance in absorption of the x-rays by different body tissues produces contrast within the image to give a 2-D representation. Example of clinical application: skeletal-examine bone structure, chest-assess lung pathology, etc.

NVIDIA cuBLAS library is a fast GPU-accelerated implementation of the standard basic linear algebra subroutines (BLAS). Using cuBLAS APIs, you can speed up your applications by deploying compute-intensive operations to a single GPU or scale up and distribute work across multi-GPU configurations efficiently.

CUDA is a parallel computing platform and application programming interface (API) model created by NVIDIA, an acronym for Compute Unified Device Architecture

NVIDIA CUDA® Deep Neural Network library (cuDNN) is a GPU-accelerated library of primitives for deep neural networks. cuDNN provides highly tuned implementations for standard routines such as forward and backward convolution, pooling, normalization, and activation layers.

Deep Learning
Deep Learning is a subfield of machine learning concerned with algorithms inspired by the structure and function of the brain called artificial neural networks. A deep neural network (DNN) is an artificial neural network (ANN) with multiple hidden layers of units between the input and output layers. Similar to shallow ANNs, DNNs can model complex non-linear relationships.

A unit of computing power equal to one billion billion floating-point operations per second. Such capacity represents a thousand-fold increase over the first petascale computer that came into operation in 2008.

Floating-point operations per second (used as a measure of computing power).

Fluoroscopy uses x-rays to allow real-time visualization of body structures buy continually emitting beams that are then captured on a screen allow for dynamic assessment of anatomy and function. Example of clinical application: barium study-gastrointestinal tract examination.

Software used to program and train deep neural network models

The ImageNet Large Scale Visual Recognition Challenge (ILSVRC) evaluates algorithms for object detection and image classification at large scale. One high level motivation is to allow researchers to compare progress in detection across a wider variety of objects -- taking advantage of the quite expensive labeling effort. Another motivation is to measure the progress of computer vision for large scale image indexing for retrieval and annotation.

Inference and Training
On a high level, working with deep neural networks is a two-stage process: First, a neural network is trained: its parameters are determined using labeled examples of inputs and desired output. Then, the network is deployed to run inference, using its previously trained parameters to classify, recognize and process unknown inputs. Find more information here.

Magnetic resonance imaging (MRI)
Magnetic resonance imaging uses magnetic radiation to visualize detailed internal structures. MRIs provide real-time, 3-D views of body organs with good soft issue contrast, making visualization of brain, spine and muscles/joints excellent. Example of clinical applications: evaluate organs, blood vessels, lymph nodes.

Mammography uses low-energy x-rays to image breast tissue for assessment of breast lesions and as a screening tool for detection of cancer. Example of clinical applications: screening, diagnostic and surveillance mammography; tumor marking.

Moore's Law
Moore's law refers to an observation made by Intel co-founder Gordon Moore in 1965. He noticed that the number of transistors per square inch on integrated circuits had doubled every year since their invention. Moore's law predicts that this trend will continue into the foreseeable future, but over the last decade has in fact decelerated, as GPU architectures have outstripped the rate of CPU performance-per-watt growth.

MXNet is a modern open-source deep learning framework used to train, and deploy deep neural networks. It is scalable, allowing for fast model training, and supports a flexible programming model and multiple languages. Amazon has chosen MXNet as its deep learning framework of choice at AWS.

Nuclear Medicine
Nuclear Medicine uses radioactive tracers, also called radiopharmaceuticals, in combination with an imaging scanner to assess bodily functions. 2 types of scans for nuclear medicine: single photon emission computed tomography (SPECT) and positron emission tomography (PET).

Positron Emission Tomography (PET)
Positron Emission Tomography (PET) uses different radiotracers which produce small particles called positrons-these photons are measured and the information is used to create a 3-D image of internal organs.

PyTorch is a python package that provides two high-level features: Tensor computation (like numpy) with strong GPU acceleration, useful in deep learning

ResNet, ResNet50
ResNets are Deep residual networks developed by Microsoft Research for Image Recognition. These networks led to 1st-place winning entries in all five main tracks of the ImageNet and COCO 2015 competitions, which covered image classification, object detection, and semantic segmentation. The robustness of ResNets has since been proven by various visual recognition tasks and by non-visual tasks involving speech and language.

Single Photon Emission Computed Tomography (SPECT)
SPECT uses gamma camera detectors to detect the gamma ray emissions from the tracers that have been given to the patient to provide 3-D images. Example of clinical applications: primarily to diagnose and track progression of heart disease, disorders of the bone, gall bladder, intestinal bleeding, etc.

TensorFlow is an open-source software library for machine learning across a range of tasks, and developed by Google to meet their needs for systems capable of building and training neural networks to detect and decipher patterns and correlations, analogous to the learning and reasoning which humans use.

Tensor Core
Tensor Cores in the Tesla V100 GPU boost the performance of matrix-matrix multiplication operations (found at the core of neural network training and inferencing) by more than 9x compared to the Pascal-based GP100 GPU. Tensor Cores and their associated data paths are custom-crafted to dramatically increase floating-point compute throughput at only modest area and power costs.

TeraFLOPS is a unit of computing power equal to one million million (1012) floating-point operations per second.

Theano is a Python library that allows you to define, optimize, and evaluate mathematical expressions involving multi-dimensional arrays efficiently, useful in deep learning.

Torch, PyTorch
Torch is an open source machine learning library, a scientific computing framework, and a script language based on the Lua programming language.[3] It provides a wide range of algorithms for deep machine learning, and uses the scripting language LuaJIT, and an underlying C implementation.

Ultrasound: Ultrasound uses high-frequency sound waves to provide cross-sectional images of the body. Example of clinical applications: visualize anatomy and pathology of organs, masses and vascular structures.

© 2022 Elliot K. Fishman, MD, FACR
All Rights Reserved.
CTisus CT Scanning CTisus CT Scanning CTisus CT Scanning CTisus CT Scanning CTisus CT Scanning CTisus CT Scanning CTisus CT Scanning CTisus CT Scanning CTisus CT Scanning CTisus CT Scanning