scholarly journals Statistical Distortion of Supervised Learning Predictions in Optical Microscopy Induced by Image Compression

2021 ◽  
Author(s):  
Enrico Pomarico ◽  
Cédric Schmidt ◽  
Florian Chays ◽  
David Nguyen ◽  
Arielle Planchette ◽  
...  
Keyword(s):  
Image Compression ◽  
Optical Microscopy ◽  
Supervised Learning ◽  
Automated Analysis ◽  
Cell Segmentation ◽  
Sensor Calibration ◽  
Predictive Uncertainty ◽  
Data Throughput ◽  
Noise Statistics ◽  
Compressed Data

Abstract The growth of data throughput in optical microscopy has triggered the extensive use of supervised learning (SL) models on compressed datasets for automated analysis. Investigating the effects of image compression on SL predictions is therefore pivotal to assess their reliability, especially for clinical use.We quantify the statistical distortions induced by compression through the comparison of predictions on compressed data to the raw predictive uncertainty, numerically estimated from the raw noise statistics measured via sensor calibration. Predictions on cell segmentation parameters are altered by up to 15% and more than 10 standard deviations after 16-to-8 bits pixel depth reduction and 10:1 JPEG compression. JPEG formats with higher compression ratios show significantly larger distortions. Interestingly, a recent metrologically accurate algorithm, offering up to 10:1 compression ratio, provides a prediction spread equivalent to that stemming from raw noise. The method described here allows to set a lower bound to the predictive uncertainty of a SL task and can be generalized to determine the statistical distortions originated from a variety of processing pipelines in AI-assisted fields.

Development and study of a method for cell separation during white blood cell segmentation on images of bone marrow preparations in information and measurement systems for diagnostics of acute leukemia

2020 ◽  
pp. 68-72
Author(s):  
V.G. Nikitaev ◽  
A.N. Pronichev ◽  
V.V. Dmitrieva ◽  
E.V. Polyakov ◽  
A.D. Samsonova ◽  
...  
Keyword(s):  
Bone Marrow ◽  
Blood Cell ◽  
Acute Leukemia ◽  
White Blood Cell ◽  
Cell Separation ◽  
Large Scale ◽  
Bone Marrow Cells ◽  
Automated Analysis ◽  
Cell Segmentation ◽  
Measurement Systems

The issues of using of information and measurement systems based on processing of digital images of microscopic preparations for solving large-scale tasks of automating the diagnosis of acute leukemia are considered. The high density of leukocyte cells in the preparation (hypercellularity) is a feature of microscopic images of bone marrow preparations. It causes the proximity of cells to eachother and their contact with the formation of conglomerates. Measuring of the characteristics of bone marrow cells in such conditions leads to unacceptable errors (more than 50%). The work is devoted to segmentation of contiguous cells in images of bone marrow preparations. A method of cell separation during white blood cell segmentation on images of bone marrow preparations under conditions of hypercellularity of the preparation has been developed. The peculiarity of the proposed method is the use of an approach to segmentation of cell images based on the watershed method with markers. Key stages of the method: the formation of initial markers and builds the lines of watershed, a threshold binarization, shading inside the outline. The parameters of the separation of contiguous cells are determined. The experiment confirmed the effectiveness of the proposed method. The relative segmentation error was 5 %. The use of the proposed method in information and measurement systems of computer microscopy for automated analysis of bone marrow preparations will help to improve the accuracy of diagnosis of acute leukemia.

scholarly journals Facetto: Combining Unsupervised and Supervised Learning for Hierarchical Phenotype Analysis in Multi-Channel Image Data

2019 ◽  
Author(s):  
Robert Krueger ◽  
Johanna Beyer ◽  
Won-Dong Jang ◽  
Nam Wook Kim ◽  
Artem Sokolov ◽  
...  
Keyword(s):  
Supervised Learning ◽  
Visual Analytics ◽  
Cancer Biology ◽  
Large Scale ◽  
Hierarchical Structures ◽  
Image Data ◽  
Automated Analysis ◽  
Cell Types ◽  
High Dimensional ◽  
Exploration Process

AbstractFacetto is a scalable visual analytics application that is used to discover single-cell phenotypes in high-dimensional multi-channel microscopy images of human tumors and tissues. Such images represent the cutting edge of digital histology and promise to revolutionize how diseases such as cancer are studied, diagnosed, and treated. Highly multiplexed tissue images are complex, comprising 109or more pixels, 60-plus channels, and millions of individual cells. This makes manual analysis challenging and error-prone. Existing automated approaches are also inadequate, in large part, because they are unable to effectively exploit the deep knowledge of human tissue biology available to anatomic pathologists. To overcome these challenges, Facetto enables a semi-automated analysis of cell types and states. It integrates unsupervised and supervised learning into the image and feature exploration process and offers tools for analytical provenance. Experts can cluster the data to discover new types of cancer and immune cells and use clustering results to train a convolutional neural network that classifies new cells accordingly. Likewise, the output of classifiers can be clustered to discover aggregate patterns and phenotype subsets. We also introduce a new hierarchical approach to keep track of analysis steps and data subsets created by users; this assists in the identification of cell types. Users can build phenotype trees and interact with the resulting hierarchical structures of both high-dimensional feature and image spaces. We report on use-cases in which domain scientists explore various large-scale fluorescence imaging datasets. We demonstrate how Facetto assists users in steering the clustering and classification process, inspecting analysis results, and gaining new scientific insights into cancer biology.

scholarly journals Image Compression with LZW (Lossless) On Different Image Formats

2018 ◽  
pp. 2153-2156
Author(s):  
Hitesh H Vandra
Keyword(s):  
Image Compression ◽  
Data Stream ◽  
Source Information ◽  
Compression Technique ◽  
Original Source ◽  
Lossless Image Compression ◽  
Source Data ◽  
Data Content ◽  
Lzw Algorithm ◽  
Compressed Data

Image compression is used to reduce bandwidth or storage requirement in image application. Mainly two types of image compression: lossy and lossless image compression. A Lossy Image Compression removes some of the source information content along with the redundancy. While the Lossless Image Compression technique the original source data is reconstructed from the compressed data by restoring the removed redundancy. The reconstructed data is an exact replica of the original source data. Many algorithms are present for lossless image compression like Huffman, rice coding, run length, LZW. LZW is referred to as a substitution or dictionary-based encoding algorithm. The algorithm builds a data dictionary of data occurring in an uncompressed data stream. Patterns of data (substrings) are identified in the data stream and are matched to entries in the dictionary. If the substring is not present in the dictionary, a code phrase is created based on the data content of the substring, and it is stored in the dictionary. The phrase is then written to the compressed output stream. In this paper we see the effect of LZW algorithm on the png, jpg, png, gif, bmp image formats.

Interactive Cell Segmentation Based on Active and Semi-Supervised Learning

2016 ◽  
Vol 35 (3) ◽  
pp. 762-777 ◽  
Author(s):  
Hang Su ◽  
Zhaozheng Yin ◽  
Seungil Huh ◽  
Takeo Kanade ◽  
Jun Zhu
Keyword(s):  
Supervised Learning ◽  
Cell Segmentation

Effective Multiple Vector Quantization for Image Compression

2007 ◽  
Vol 11 (10) ◽  
pp. 1189-1196
Author(s):  
Noritaka Shigei ◽  
◽  
Hiromi Miyajima ◽  
Michiharu Maeda ◽  
Lixin Ma ◽  
...  
Keyword(s):  
Neural Network ◽  
Image Compression ◽  
Image Restoration ◽  
Vector Quantization ◽  
High Quality ◽  
Index Coding ◽  
Network Algorithm ◽  
Neural Network Algorithm ◽  
Compressed Data

Multiple-VQ methods generate multiple independent codebooks to compress an image by using a neural network algorithm. In the image restoration, the methods restore low quality images from the multiple codebooks, and then combine the low quality ones into a high quality one. However, the naive implementation of these methods increases the compressed data size too much. This paper proposes two improving techniques to this problem: “index inference” and “ranking based index coding.” It is shown that index inference and ranking based index coding are effective for smaller and larger codebook sizes, respectively.

Stabilization of the compressed data formation rate in hierarchical image compression

2007 ◽  
Vol 17 (1) ◽  
pp. 79-81 ◽  
Author(s):  
M. V. Gashnikov ◽  
N. I. Glumov ◽  
V. V. Sergeyev
Keyword(s):  
Image Compression ◽  
Formation Rate ◽  
Compressed Data

scholarly journals Supervised‐learning‐based algorithm for color image compression

ETRI Journal ◽  
2020 ◽  
Vol 42 (2) ◽  
pp. 258-271
Author(s):  
Xue‐Dong Liu ◽  
Meng‐Yue Wang ◽  
Ji‐Ming Sa
Keyword(s):  
Image Compression ◽  
Supervised Learning ◽  
Color Image ◽  
Color Image Compression
Sign in / Sign up

Export Citation Format

Share Document