Low-cost thermal imaging with machine learning for non-invasive diagnosis and therapeutic monitoring of pneumonia

Abstract

Rapid screening and early treatment of lung infection are essential for effective control of many epidemics such as Coronavirus Disease 2019 (COVID-19). Recent studies have demonstrated the potential correlation between lung infection and the change of back skin temperature distribution. Based on these findings, we propose to use low-cost, portable and rapid thermal imaging in combination with image-processing algorithms and machine learning analysis for non-invasive and safe detection of pneumonia. The proposed method was tested in 69 subjects (30 normal adults, 11 cases of fever without pneumonia, 19 cases of general pneumonia and 9 cases of COVID-19) where both RGB and thermal images were acquired from the back of each subject. The acquired images were processed automatically in order to extract multiple location and shape features that distinguish normal subjects from pneumonia patients at a high accuracy of 93 $\%$ . Furthermore, daily assessment of two pneumonia patients by the proposed method accurately predicted the clinical outcomes, coincident with those of laboratory tests. Our pilot study demonstrated the technical feasibility of portable and intelligent thermal imaging for screening and therapeutic assessment of pneumonia. The method can be potentially implemented in under-resourced regions for more effective control of respiratory epidemics.

Keywords: Diagnosis; Machine learning; Pneumonia; Therapeutic monitoring; Thermal imaging.

Fig. 1

Overview of the study.

Fig. 2

(a) schematic diagram of thermal imaging system for pneumonia screening. (b) procedures of feature selections.

Fig. 3

Typical examples of thermal imaging results of one normal person, one pneumonia patient and one fever patient without pneumonia. (a) thermal imaging result of one normal person; (b) thermal imaging result of one pneumonia patient shows temperature increases (red or white areas) in both right and left lung; (c) thermal imaging result of one fever patient without pneumonia.

Fig. 4

Typical examples of one recovered pneumonia patient after 3-days therapy. The thermal imaging images were taken from day 1 to day 3.

Fig. 5

Typical examples of one unrecovered pneumonia patient after 4-days therapy. The thermal imaging images were taken from day 1 to day 4.

Fig. 6

Plots of pneumonia probability by SVM model based on normal and pneumonia. (a) Plot of pneumonia probability of all subjects. (b) Plot of pneumonia probability of one recovered pneumonia patient after a 3-day therapy. (c) Plot of pneumonia probability of one unrecovered pneumonia patient after a 4-day therapy.

Fig. 7

Plots of pneumonia probability by SVM model based on pneumonia positive and pneumonia negative. (a) Plot of pneumonia probability of all subjects. (b) Plot of pneumonia probability of one recovered pneumonia patient after 3-days therapy. (c) Plot of pneumonia probability of one unrecovered pneumonia patient after 4-days therapy.

Fig. 8

Plots of pneumonia probability by SVM model based on normal, fever and pneumonia positive. (a) Plot of pneumonia probability of all subjects. (b) Plot of pneumonia probability of one recovered pneumonia patient after 3-days therapy. (c) Plot of pneumonia probability of one unrecovered pneumonia patient after 4-days therapy.

Fig. 9

Ranking of top feature importance. The feature importance is evaluated based on SVM models.