A comparative study of source-finding techniques in HI emission line cubes using SoFiA, MTObjects, and supervised deep learning

Abstract

The 21 cm spectral line emission of atomic neutral hydrogen (HI) is one of the primary wavelengths observed in radio astronomy. However, the signal is intrinsically faint and the HI content of galaxies depends on the cosmic environment, requiring large survey volumes and survey depth to investigate the HI Universe. As the amount of data coming from these surveys continues to increase with technological improvements, so does the need for automatic techniques for identifying and characterising HI sources while considering the tradeoff between completeness and purity. This study aimed to find the optimal pipeline for finding and masking the most sources with the best mask quality and the fewest artefacts in 3D neutral hydrogen cubes. Various existing methods were explored in an attempt to create a pipeline to optimally identify and mask the sources in 3D neutral hydrogen 21 cm spectral line data cubes. Two traditional source-finding methods were tested, SoFiA and MTObjects, as well as a new supervised deep learning approach, in which a 3D convolutional neural network architecture, known as V-Net was used. These three source-finding methods were further improved by adding a classical machine learning classifier as a post-processing step to remove false positive detections. The pipelines were tested on HI data cubes from the Westerbork Synthesis Radio Telescope with additional inserted mock galaxies. SoFiA combined with a random forest classifier provided the best results, with the V-Net-random forest combination a close second. We suspect this is due to the fact that there are many more mock sources in the training set than real sources. There is, therefore, room to improve the quality of the V-Net network with better-labelled data such that it can potentially outperform SoFiA.

References in corpus (29)

• Adam: A Method for Stochastic Optimization
• Scikit-learn: Machine Learning in Python
• YOLOv4: Optimal Speed and Accuracy of Object Detection
• An overview of gradient descent optimization algorithms
• Fully Convolutional Networks for Semantic Segmentation
• Galaxy Zoo : Morphologies derived from visual inspection of galaxies from the Sloan Digital Sky Survey
• k-Nearest Neighbour Classifiers: 2nd Edition (with Python examples)
• The Pan-STARRS1 Database and Data Products
• Automatic Liver and Lesion Segmentation in CT Using Cascaded Fully Convolutional Neural Networks and 3D Conditional Random Fields
• Fundamental Physics with the Square Kilometre Array
• SoFiA: a flexible source finder for 3D spectral line data
• SoFiA 2 -- An automated, parallel HI source finding pipeline for the WALLABY survey
• Classifying Radio Galaxies with Convolutional Neural Network
• Optimising Automatic Morphological Classification of Galaxies with Machine Learning and Deep Learning using Dark Energy Survey Imaging
• Tracing HI Beyond the Local Universe
• Wide Field Imaging. I. Applications of Neural Networks to object detection and star/galaxy classification
• The FIRST Classifier: Compact and Extended Radio Galaxy Classification using Deep Convolutional Neural Networks
• Using negative detections to estimate source finder reliability
• The DiskMass Survey. X. Radio synthesis imaging of spiral galaxies
• Deep Learning Based Detection of Cosmological Diffuse Radio Sources
• Comparison of potential ASKAP HI survey source finders
• Optimising and comparing source extraction tools using objective segmentation quality criteria
• Data-Efficient Classification of Radio Galaxies
• The Characterised Noise Hi source finder: Detecting Hi galaxies using a novel implementation of matched filtering
• The Cosmic Large-Scale Structure in X-rays (CLASSIX) cluster survey IV: Superclusters in the local Universe at z <= 0.03
• The Cosmic Large-Scale Structure in X-rays (CLASSIX) Cluster Survey III: The Perseus-Pisces supercluster and the Southern Great Wall as traced by X-ray luminous galaxy clusters
• Statistical analysis of undetected point sources in cosmic microwave background maps
• The role of 3-D interactive visualization in blind surveys of HI in galaxies
• Image Segmentation using Unsupervised Watershed Algorithm with an Over-segmentation Reduction Technique

Cited by in corpus (2)

• SKA Science Data Challenge 2: analysis and results
• WALLABY Pilot Survey: HI source-finding with a machine learning framework