The Accuracy of Electrical Impedance Tomography for Breast Cancer Detection: A Systematic Review and Meta-Analysis

Introduction Incidence of breast cancer (BC) in 2020 is about 2.26 million new cases. It is the first common cancer accounting for 11.7% of all cancer worldwide. Disease complications and the mortality rate of breast cancer are highly dependent on the early diagnosis. Therefore, novel human breast-imaging techniques play an important role in minimizing the breast cancer morbidity and mortality rate. Electrical impedance tomography (EIT) is a noninvasive technique to image the breast using the electrical impedance behavior of the body tissues. Objectives The aims of this manuscript are as follows: (1) a comprehensive investigation of the accuracy of EIT for breast cancer diagnosis through searching pieces of evidence in the valid databases and (2) meta-analyses of the results. Methods The systematic search was performed in the electronic databases including PubMed, Web of Science, EMBASE, Science Direct, ProQuest, Scopus, and Google Scholar without time and language limitation until January 2021. Search terms were “EIT” and “Breast Cancer” with their synonyms. Relevant studies were included based on PRISMA and study objectives. Quality of studies and risk of bias were performed by QUADAS-2 tools. Then, relevant data were extracted in Excel form. The hierarchical/bivariate meta-analysis was performed with “metandi” package for the ROC plot of sensitivity and specificity. Forest plot of the Accuracy index and double arcsine transformations was applied to stabilize the variance. The heterogeneity of the studies was evaluated by the forest plots, χ2 test (assuming a significance at the a-level of 10%), and the I2 statistic for the Accuracy index. Results A total of 4027 articles were found. Finally, 12 of which met our criteria. Overall, these articles included studies of 5487 breast cancer patients. EIT had an overall pooled sensitivity and specificity of 75.88% (95% CI, 61.92% to 85.89%) and 82.04% (95% CI, 69.72% to 90.06%), respectively. The pooled diagnostic odds ratio was 14.37 (95% CI, 6.22% to 33.20%), and the pooled effect of accuracy was 0.79 with 95% CI (0.73, 0.83). Conclusions This study showed that EIT can be used as a useful method alongside mammography. EIT sensitivity could not be compared with the sensitivity of MRI, but in terms of specificity, it can be considered as a new method that probably can get more attention. Furthermore, large-scale studies will be needed to support the evidence.


Introduction
Incidence of breast cancer (BC) in 2020 is about 2.26 million new cases. It is the first common cancer accounting for 11.7% of all cancer worldwide [1]. It is the leading cause of cancer death in most countries [2]. e mortality rate of breast cancer varies from region to region. Compared to the USA and Europe, some Asian countries such as Japan and China have lower mortality rates due to breast cancer. Breast cancer mortality rate is higher than the mortality rate of colorectal cancer (8%) and lower than the mortality rate of lung cancer [3].
Similar to many other forms of cancers, breast cancer is the result of several environmental and hereditary factors. High-fat diets, alcohol intake, tobaccos, hormones, radiation, age, sex, and childbearing are identified as the primary risk factors [4,5]. Statistical data show that many women with breast tumors refer to the hospital only after they feel a quite large lump. At that stage of the disease, a biopsy is required to determine whether the tumor is cancerous (malignant) or noncancerous (benign). For this purpose, clinicians often perform a fine-needle aspiration or a core biopsy [6]. e type of the tumor determines the subsequent treatments. While the core biopsy is a reliable technique to characterize the tumors, it is not hazard-free. e process requires taking out a considerable amount of tissue from the body at several locations. ere are many reports on cancerous tumor adhering to the tip of the biopsy needle and introducing the disease to the upper cutaneous layers [7]. e early diagnosis of breast cancer plays a crucial role in minimizing the mortality rate of the disease. Magnetic resonance imaging (MRI), computed tomography (CT) scan, ultrasound imaging, and mammography are current techniques implemented for diagnostic imaging. Among these techniques, mammography became a standard technique for breast imaging and early-stage detection of carcinomas in the breast [8,9]. Ultrasound imaging is also employed along with mammography to differentiate suspicious breast lesions [10]. While the mammography technique offers invaluable advantages to decrease mortality [11], it has some drawbacks as well. e disadvantage is that it applies ionizing radiation, which is associated with patient discomfort because of breast tissue compression [12]. In addition to the radiation exposure issues, the mammogram may result in false-positive results. It also has a sensitivity of lower than 100% [13]. Also, MRI has significant disadvantages as well, including its high cost, variability in performance, and moderate specificity [14], and application of CT scan has some limitations, such as portability, intermittent use, and X-ray exposure [15].
Hence, there is a need for the development of reliable noninvasive and nonhazardous imaging techniques for the detection and characterization of breast cancer tumors.
Electrical impedance tomography (EIT) is a noninvasive technique to image the human breast using electrical impedance characteristics of the tissues [16]. In this imaging technique, an array of external electrodes is employed to extract and reconstruct an image of the inner part of a conductive object. e value of the electrical impedance in human tissues depends on the tissue's electrical storage potential. e electrical impedance is different in the normal tissues and pathologically changed tissues [17]. e variation of cellular water content, variation of extracellular fluids, packing density, and differences in the orientation of the cells are the main reasons for a significant increase in capacitance and conductance of malignant tumors, which results in lower impedance [18].
EIT is a rapid, compact, and inexpensive imaging solution [19][20][21]. e EIT technique can be employed with almost no age limitations. It can also be used on pregnant women because no ionizing radiation is emitted by the scanning device [16].
Several studies that have been performed previously focused on investigating the accuracy of EIT systems in the detection of breast cancer. ese studies did not show comprehensive results in this field. However, to the best of the authors' knowledge, there are no comprehensive metaanalyses on this matter. On the other hand, from the clinical point of view, rare evidence is available for using this method for detecting breast cancer. So, there is still a gap between clinical use and research in this field. erefore, the aim of this manuscript is as follows: (1) a comprehensive investigation of the accuracy of the electrical impedance tomography technique for breast cancer diagnosis through searching evidence in the valid databases and (2) meta-analyses of the results. e reference lists of relevant primary studies, reviews, and key journals for additional studies were also searched.

Study Selection Process.
is systematic review and meta-analysis were conducted based on the PRISMA statement. EndNote software manager was used to manage the references. After removing duplications, two authors independently reviewed references based on the title and abstract. We excluded narrative reviews, opinion pieces, letters, and any other publications lacking primary data. Any disagreements between the investigators regarding the inclusion of articles were discussed and resolved with a third reviewer.

Inclusion Criteria.
Articles were included based on the following inclusion criteria: (1) Observational and interventional studies that assessed using the EIT to diagnose breast cancer 2 e Breast Journal (2) Published articles in any language with full English abstracts were considered (3) And explicitly calculated performance of the EII through performance metrics 2.4. Data Items. Two authors independently collected the data from articles and any conflicts were solved by the third author. e relevant data were extracted and collected in Excel form. Data items were author, year, study design, objectives, specificity, sensitivity, PPV, NPV, ACC, number of patients, age, TP, FP, FN, and TN.

Risk of Bias (Quality)
Assessment. e quality of methodological of primary studies was assessed by the modified QUADAS-2 quality assessment tool which was developed by Whiting [22]. QUADAS is an 11-item generic tool that was developed specifically for implementation in diagnostic test accuracy reviews. QUADAS was developed using a formal consensus method informed by empirical evidence.

Effect Measures.
In this study, the reference test was EIT and the gold standard was mammography. Specificity, sensitivity, ACC, and diagnostic odds ratio were the variables for the outcome measure.
We conducted Cook's distance analysis as a measure of the influence of a study on the model.

Statistical Analysis.
e statistical analyses were performed by using STATA V.13 statistical software (Stata Corp. 2013. Stata Statistical Software: Release 13. College Station, TX: Stata Corp LP). e hierarchical/bivariate metaanalysis was performed with the "metandi" package for the ROC plot of sensitivity and specificity. Patient frequencies within extracted 2 × 2 data tables were implemented to generate the forest plot of the Accuracy index that was applied using the "metaprop" command, and double arcsine transformations were applied to stabilize the variance. Every study that did not have enough outcome variables was eliminated.
We evaluated the heterogeneity among the primary studies by the forest plots, χ2 test (assuming a significance at the a-level of 10%), and the I 2 statistic for the Accuracy index.

Study Characteristics.
A total of 4027 articles were initially retrieved from the electronic databases. After removing the duplicates, the remaining 3544 studies were screened. A total of 3435 studies were excluded. Among 109 remaining papers, 88 were excluded because their methods were not available or the reference test was different or it did not provide relevant data. Finally, 12 articles were selected for inclusion in this study. Figure 1 illustrates the selection process of the articles based on the PRISMA flow diagram. e 12 selected studies included data for 5487 patients with breast cancer. ese studies were conducted between 1999 and 2017. Most of these studies were conducted in Germany, China, and the USA.
e main characteristics of each study and their sensitivity and specificity are summarized in Table 1.

Reporting Bias and Quality
Assessment. According to QUADAS-2, the quality of the studies was evaluated. Each study was evaluated by two independent investigators. All disagreements were resolved by consensus. e quality of the included studies based on the QUADAS-2 tool showed a high risk of bias (Table 2). Seven articles had a high risk of bias, and 5 articles had a low risk of bias.

Meta-Analysis.
e summary ROC curve presents the summary estimates for EIT ( Figure 2). Table 3 illustrates the bivariate and the hierarchical summary receiver operating characteristic (HSROC) parameter estimates along with their standard errors and approximate 95% confidence intervals. Pooled sensitivity and specificity were 75.88 and 82.04 with the standard error of 6.16 and 5.14, respectively. e pooled diagnostic odds ratio (DOR) was 14.37 with a 6.14 standard deviation. Figure 2 shows the summary ROC plot of sensitivity and specificity of EIT. is diagram is a graphical representation of a fitted model for simultaneous meta-analysis of Se and Sp on a ROC diagram. Each small circle of information is specific to a study, and the size of which is related to the weight assigned to it in the meta-analysis. e confidence interval and the prediction intervals are also marked with dashes. e pooled sensitivity and feature are also marked with a red dot. is graph uses the hierarchical summary receiver operating characteristic (HSROC) method to estimate results.
Cook's distance analysis was conducted as a measure of the influence of a study on the model parameters [34]. e results indicate that two studies (ID numbers 8 and 12) had the most influence on the parameters (Figure 3). Table 4 shows the results of the pooled effect of accuracy of 0.79 with 95% CI (0.73, 0.83) and P value � <0.01.
Significant heterogeneity was found between the studies (I 2 � 93.26% and P value <0.01). e forest plot is shown in Figure 4. Due to the heterogeneity of the studies, the data were combined using a random-effect model. Different comparisons with different diagnostic methods may cause heterogeneity in our study.
Funnel plot was used to examine the publication bias, and the distribution of points in the two halves of the drawn triangle is discussed. Each point of the number is related to a study, which if these points are drawn relatively uniformly in all parts of the triangle, indicates the absence of publication bias. Finally, no significant publication bias was found through the funnel plot (P value � 0.108) ( Figure 5).        82.04 ± 5.14, respectively) and a DOR of 14.37 ± 6.14. e results of this meta-analysis revealed overall high sensitivity and specificity. Based on our knowledge, our study is the first systematic review and meta-analysis to investigate the accuracy of the EIT for breast cancer detection. In this field, only one systematic review was performed by Zain et al. in 2015 and their review did not reach the meta-analysis. eir review showed that the range sensitivity of the EIT system to the human breast was between 17% and 94.6%. e range of specificity was between 49% and 97.1%. e accuracy (ACC) of EIT was between 69% and 80.5% [35].
Although MRI is an accepted tool in detecting tumors, breast MRI is expensive and can only diagnose breast lesions in certain clinical situations [39]. As such, alternative devices could be favorable. EIT is a radiation-free and noninvasive method that can be employed as an inexpensive, portable, and radiation-free technique for imaging [20,21]. Hence, it is a great candidate for bedside diagnosis and intraoperative imaging applications [40]. However, EIT was not acceptable for clinical diagnosis until now, and more research is needed for its development.
Several studies were conducted using meta-analyses for mammography accuracy stand-alone or in comparison with other imaging techniques. e range of pooled sensitivity of 75-97% and the specificity of 66-96% with 95% CI [36,38,41] was reported. e sensitivity of 75% and specificity of 82% of our study showed that EIT can be used as a useful method alongside mammography.
In comparison with ultrasound or mammography, EIT is less effective due to its low specificity [27]. It appears that age is an important determinative factor in this regard. As seen in studies [29,30], EIT may have more efficiency for people who are 30-45 years old. Breast cancers in young women are usually diagnosed at a more advanced stage, requiring more serious and expensive treatments with a reduction in the survivorship rate and the quality of life [42]. is late diagnosis is due to the lack of mammography evaluation for people under 40 years old. In addition, small and nonpalpable probably benign lesions remain unnoticed with either clinical, self-breast examination, or screening surveillance until they have grown sufficiently large enough that the woman herself detects a palpable mass [42,43].   e Breast Journal erefore, for people under 40, the use of low-cost modalities with acceptable sensitivity and specificity at the initial stages of the disease is required to reduce the rate of unnecessary biopsies [42].
Screening surveillance in the probably benign lesions is more cost-effective than surgical intervention and prevents unnecessary biopsies [44]. However, despite all efforts for follow-ups, approximately 30 to 47% of participants did not complete the screening surveillance [45,46]. EIT has demonstrated good results in detecting nonpalpable breast cancer as a safe diagnostic tool in younger women [42].
Subgroup analysis can provide more specific information for diagnosis in special groups. In this study, we did not perform subgroup analysis because of limited initial information in selected studies. Respectively, discrimination between patients of different ages was not possible, for example, in patients under 40, the sensitivity of mammography to detect cancer is decreased [47].
Different quality of articles may affect the result of metaanalysis. However, it showed that the use of EITcan be useful because of its sensitivity in detecting cancer in cases that were not detected by mammography and ultrasound [48].

Limitations of the Study
(1) Unfortunately, it is not possible to analyze the histopathology of patients to determine the efficiency of EIT in detecting the type of different mass in young women (2) Different comparisons with different diagnostic methods may cause heterogeneity in our study (3) Insufficient information and unclear mean age in different groups in articles

Conclusion
is paper demonstrated an acceptable accuracy of EIT for breast cancer detection. However, it is concluded that ETL is more considered in the research and clinical application was not very clear yet. is study showed that EIT can be used as a useful method alongside mammography. EIT sensitivity could not be compared with MRI, but in terms of specificity, it can be considered as a new method that probably can get more attention. Furthermore, large-scale studies will be needed to support the evidence.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.