Sistem za avtomatsko detekcijo artefaktov na mamografskih slikah : magistrsko delo

Iskra, Aljaž (author), Žibert, Janez (mentor)

Uvod: Odkrivanje raznovrstnih artefaktov na slikah zaradi posledic različnih dejavnikov je ključnega pomena pri nadzoru kakovosti delovanja mamografskih slikovnih sistemov, ki jo vsakodnevno izvajajo radiološki inženirji. V okviru magistrskega dela smo obravnavali temeljne korake razvoja avtomatskega sistema za detekcijo artefaktov: zajem in označevanje homogenih slik, obdelavo slik za razvoj značilk, izpeljavo in izbiro lastnosti slik, ki merijo verjetnost določene vrste popačenja na sliki ter modeliranje s pomočjo različnih klasifikacijskih modelov in njihovo evalvacijo. Namen: Cilj raziskave je bil razvoj sistema za avtomatsko detekcijo artefaktov na mamografskih slikah, ki bi se lahko uporabljal za analizo delovanja mamografskih slikovnih sistemov v okviru vsakodnevne kontrole kvalitete s pomočjo slikanja homogenih fantomov. Metode dela: Definirali smo protokol za zajem homogenih mamografskih slik, razvili lastnosti za izpeljavo mer kakovosti in testirali različne klasifikacijske metode za detekcijo artefaktov. Sistem je bil razvit z namenom razpoznavanja in zaznavanja artefaktov nehomogenosti kontrasta, ghosting artefaktov, linijskih artefaktov, mrtvih slikovnih elementov ter napak pri samem procesu zajema slike in ostalih artefaktov, ki se najpogosteje pojavljajo na mamografskih in drugih digitalnih rentgenskih slikah. Rezultati: Zajeli smo 542 homogenih slik na 7 različnih mamografskih slikovnih sistemih, na katerih je bilo v 101 primeru prisotnih artefaktov nehomogenosti kontrasta, 53 ghosting artefaktov, 52 linijskih artefaktov, 18 mrtvih slikovnih elementov in 51 primerov drugih napak v povezavi z zajemom slike. Detekcijo artefaktov smo izvedli s pomočjo razvoja različnih značilk in tehnik modeliranja. Izpeljali smo devet različnih mer, ki preučujejo homogenost, teksturo slike, povprečno razliko sivinskih nivojev na izbranem področju, prisotnost linij in mrtvih slikovnih elementov. S pomočjo naslednjih petih klasifikacijskih modelov smo testirali različne kombinacije značilk: z logistično regresijo, metodo naivnega Bayesovega klasifikatorja, metodo odločitvenih dreves, metodo naključnih gozdov in metodo nevronske mreže z večplastnimi perceptroni. Po evalvaciji z navzkrižno validacijo izpusti enega (z izračunom AUC v ROC analizi) smo v 98.6% uspeli zaznati nehomogenosti kontrasta, v 92.2% ghosting artefakte, v 95.1% linijske artefakte, v 100% mrtve slikovne elemente in v 97.8% druga popačenja. Razprava in zaključek: Najboljše rezultate je dala metoda naključnih gozdov, pri kateri smo v klasifikacijo vključili vseh devet mer kakovosti slik, ali v nekaterih primerih samo specifično razvito mero za zaznavo posameznega artefakta.

• Language: Slovenian
• Keywords: detekcija artefaktov, mamografija, obdelava slike, strojno učenje
• Work type: Master's thesis/paper
• Organization: ZF - Faculty of Health Sciences
• Year: 2019
• PID: 20.500.12556/RUL-111217
• COBISS.SI-ID: 5698667
• Publication date in RUL: 26.09.2019

Secondary language

• Language: English
• Title: Automated artifact detection in uniform mammography images : master thesis

Abstract

Introduction: Detection of various artifacts in images, caused by many factors, is essential for evaluating quality control of mammography units that is daily performed by the radiographer. The paper addresses all the important issues in the process of building an automatic system for artifact detection: image acquisition and labeling, image preprocessing for feature extraction, feature extraction and selection, and classification modeling with evaluation. Purpose: The aim of this research was to build a system for the automatic detection of artifacts in mammographic images that can be used in practice as an analysis tool for the daily quality control of flat-field homogenous images in mammography systems. Methods: We defined the procedures for image acquisition, developed measures for feature extraction and tested different classification methods for artifact detection. Our system was designed to recognize artifacts on the basis of contrast non-uniformity, ghosting, lines, dead pixels and acquisition errors, which are the most frequently observed artifacts in mammographic and other DR radiographic images. Results: We acquired 542 mammographic images produced with 7 different mammography units, of which 101 images have non-uniform contrast, 53 have ghosting artifacts, 52 have line artifacts, 18 have dead pixels and 51 have artifacts due to acquisition errors. The artifact detection was performed using different features and different modeling techniques. We designed nine different features, which measure the uniformity, texture information, average gray-level segment difference and the presence of lines and dead pixels. Different combinations of features were tested with five classification methods: logistic regression, na?ve Bayes, decision tree, random forest and neural network with a multilayer perceptron. A leave-one-out cross validation yielded the best overall results (measured by the AUC in the ROC analysis) of 98.6% for detecting non-uniform contrast, 92.2% for detecting ghosting artifacts, 95.1% for line artifacts, 100% for detecting images with dead pixels and 97.8% for detecting artifacts due to acquisition errors. Discussion and conclusion: The best overall results were achieved by using the random forest for classification together with all nine features or in some cases just with the specially designed features for the specific artifact.

• Keywords: artifact detection, mammography, image processing, machine learning