AI Applications in Cardiology, Radiology, CT and MR Images

Abstract

AI is becoming irrefutable in many fields, but in the field of medical it has become very important. It has many applications in Health and Bio-medical which is currently being deployed and also significant research is still going on. It mainly focuses on diagnoses of diseases through image processing. Few papers about uses of AI in cardiology and radiology, based on ECG and CT and MR images, have been reviewed to extract some useful insights.

Index Terms: ECG, CT, MRI, AUC, Machine Learning

Radiography

Different techniques have been implemented for the diagnosis of different chest diseases such as fibrosis, edema and Pneumothoraces etc. As for clinicians, if the currently available deep learning models get implemented, it will be very useful for clinicians and hence they will be having very less role in disease classification and diagnosis and will be able to give due attention to patient’s treatment [12].

Important clinical abnormalities were detected on the radiography of chest and its performance was as good as that of clinicians [1]. A deep learning model was employed to detect significant Pneumothoraces where its performance was as good as of having AUC 0.94-0.96[2]. Pneumonia detection turned to be accurate on the same site -up to 0.94 AUC- while well less on any alien site data with AUC of 0.75-0.89[3] Quantification of visceral and subcutaneous fat from mouse abdomen for early stage was done [10]. Radiology professionals, researchers and clinicians also use it for different purposes like, classifying labeled images for gaining confidence in their decision, handing of the increasing data easily and many more [11].

CT and MR Images

Theses images turned to have, so far, shown much significance in AI applications in Medical. It has many application like; Detection, diagnosis, staging and sub-classification of different severe diseases. Deep learning approach was used to detect knee abnormalities using its MR images [4]. Detection of cerebral aneurysms was done on MR angiography with false/positive findings of 0.94/ 2.90 on high sensitivity models [5]. But this model gives very different results for low sensitivity results.

Liver masses are classified into five categories using CT images. These categories are “heptacellular as A and liver cyst as E” [6]. Another model was employed to stage liver fibrosis with AUC of 0.85 [7]. The genomic status of gliomas was also estimated by a deep learning model which was trained on MR images that can predict isocitrate dehydrogenase-1 mutation status and O6-methylguanine-DNA methyltransferase promotor methylation status with respective accuracy of 0.94 and 0.83 [8]. Cancer patient prognosis can also be predicted using MR and CT images and also mortality risk groups have categorized from low-high [9].

Cardiology

The heart rate is governed by different factors. It often increases with fear, anxiety or physical activity. However if such condition occurs due to abnormality in the hearts conduction system, we talk about arrhythmia. Such conditions might be caused by an obstruction or short circuit in the electrical pathways of the heart, causing improper pumping of the blood. All such abnormalities are termed Cardiac Arrhythmia or Dysrhythmia. Although most arrhythmias are benign, some of them are so serious in nature that they may cause sudden death if left unattended. These different arrhythmias are identified by clinicians/ cardiologists from their respective ECG reports.

Electrodes are placed on the body of the patient for the collection of ECG report which is then further amplified by some EC (Electronic Circuit). Leads of these electrodes have their standard positions. ECG patterns are recorded. A physician’s job is to infer some information from any change in the pattern of ECG. Each cycle of a heartbeat is represented with an ECG complex. This information is given to us by an ECG complex which further gains insights from the cycle of heartbeat which shows different activities of the heart. The main points on an ECG complex, also called the fiducially points are, (P, Q, R, S and T). In the right atrium of the heart, Sino-Atrial is located where originated an electrical impulse. This originated pulse spreads in the whole of the Atria.

Depolarization of the atrial muscle causes P wave of the ECG. The P wave is generated in response to the impulse and also has a +ve peak.

Following steps are performed in the classification task: a) First of all, we had collected a database with already having arrhythmia type known. b) For the detection of the main points of ECG i.e. P, Q, R, S, T; a new algorithm is being proposed. c) From these detected points, important feature parameters are extracted. d) Using these parameters, the computationally intelligent networks are being trained for the detection of different arrhythmia types. e) And finally, arrhythmia is classified into different types.

For detecting cardiac arrhythmias, a CGPANN trained networked algorithm is developed that first detects the ECG fiducially points, and then from these points it calculates the different morphological parameters [13].

In another work by Dawes et al. [14] pulmonary hypertension has been predicted using MR images systolic cardiac motion with high accuracy. In this experiment data was collected from 250 patients.

Phenotypic classification was done using a deep learning model. This was specifically about phenotypes with pre-served ejection fraction and heart failure [15].

Conclusion

In a nutshell, it can be said that AI and Machine Learning has so far outperformed in the field of clinical Radiology and Cardiology. This will turn very friendly for the professionals and other beneficiaries. Significant research has so far been carried out and many more has to come.

References

1. Rajpurkar P, Irvin J, Ball RL, Zhu K, Yang B, Mehta H, et al. Deep learning for chest radiograph diagnosis: A retrospective comparison of CheXNeXt to practicing radiologists. PLoS Med. 2018;15(11): e1002686. https://doi.org/10.1371/journal.pmed.1002686
2. Taylor AG, Mielke C, Mongan J. Automated detection of clinically-significant pneumothorax on frontal chest X-rays using deep convolutional neural networks. PLoS Med. 2018;15(11):e1002697. https://doi.org/10.1371/journal.pmed.1002697
3. Zech JR, Badgeley MA, Liu M, Costa AB, Titano JJ, Oermann EK. Variable generalization performance of a deep learning model to detect pneumonia in chest radiographs: A cross-sectional study. PLoS Med. 2018;15(11):e1002683. https://doi.org/10.1371/journal.pmed.1002683
4. Bien N, Rajpurkar P, Ball RL, Irvin J, Park AK, Jones E, et al. AI-assisted diagnosis for knee MR: Development and retrospective validation. PLoS Med. 2018;15(11):e1002699. https://doi.org/10.1371/journal.pmed.1002699
5. Nakao T, Hanaoka S, Nomura Y, Sato I, Nemoto M, Miki S, et al. Deep neural network-based computer-assisted detection of cerebral aneurysms in MR angiography. J Magn Reson Imaging 2018; 47 (4):948–953. https://doi.org/10.1002/jmri.25842 PMID: 28836310
6. Yasaka K, Akai H, Abe O, Kiryu S. Deep Learning with Convolutional Neural Network for Differentiation of Liver Masses at Dynamic Contrast-enhanced CT: A Preliminary Study. Radiology 2018; 286(3):887–896. https://doi.org/10.1148/radiol.2017170706 PMID: 29059036
7. Yasaka K, Akai H, Kunimatsu A, Abe O, Kiryu S. Liver Fibrosis: Deep Convolutional Neural Network for Staging by Using Gadoxetic Acid-enhanced Hepatobiliary Phase MR Images. Radiology 2018; 287 (1):146–155. https://doi.org/10.1148/radiol.2017171928 PMID: 29239710
8. Chang P, Grinband J, Weinberg BD, Bardis M, Khy M, Cadena G, et al. Deep-Learning Convolutional Neural Networks Accurately Classify Genetic Mutations in Gliomas. AJNR Am J Neuroradiol 2018; 39 (7):1201–1207. https://doi.org/10.3174/ajnr.A5667 PMID: 29748206
9. Hosny A, Parmar C, Coroller T, Grossmann P, Zeleznik R, Kumar A, et al. Deep learning for lung cancer prognostication: A retrospective multi-cohort radiomics study. PLoS Med. 2018;15(11):e1002711 https://doi.org/10.1371/journal.pmed.1002711
10. Grainger AT, Tustison NJ, Qing K, Roy R, Berr SS, Shi W. Deep learning-based quantification of abdominal fat on magnetic resonance images. PLoS ONE 2018; 13(9):e0204071. https://doi.org/10.1371/journal.pone.0204071 PMID: 30235253
11. Liu F, Jang H, Kijowski R, Bradshaw T, McMillan AB. Deep Learning MR Imaging-based Attenuation Correction for PET/MR Imaging. Radiology 2018; 286(2):676–684. https://doi.org/10.1148/radiol.2017170700 PMID: 28925823
12. Yasaka K, Abe O (2018) Deep learning and artificial intelligence in radiology: Current applications and future directions. PLoS Med 1. 15(11): e1002707. https://doi.org/10.1371/journal.pmed.1002707
13. Cardiac Arrhythmia Classification Using Cartesian Genetic Programming Evolved Artificial Neural Network/Masood Ahmad Arbab, Gul Muhammad Khan and Ali Mahmud Sahibzada/Exp Clin Cardiol Vol 20 Issue9 pages 5334-5348 / 2014
14. Dawes TJW, de Marvao A, Shi W, Fletcher T, Watson GMJ, Wharton J, Rhodes CJ, Howard LSGE, Gibbs JSR, Rueckert D, Cook SA, Wilkins MR, O’Regan DP. Machine learning of three-dimensional right ventricular motion enables outcome predictionin pulmonary hypertension: a cardiacMR imaging study. Radiology. 2017;283(2):381–90. 4.
15. Shah SJ, Katz DH, Selvaraj S, Burke MA, Yancy CW, Gheorghiade M, Bonow RO, Huang CC, Deo RC. Phenomapping for novel classification of heart failure with preserved ejection fraction. Circulation. 2015;131(3):269–79.
16. D. Bonderman, Artificial intelligence in cardiology in Wien Klin Wochenschr The Central European Journal of Medical, 8 Sep. 2017.