Globally, breast cancer is the most prevalent cancer among women, with a recorded incidence of 2.2 million cases in 2020 [1–2]. This incidence surpasses that of colorectal cancer, the second-highest cancer incidence among women, by nearly threefold [2]. Beyond its high incidence, breast cancer claims the foremost position as the leading cause of cancer-related deaths in women, resulting in approximately 685,000 lives lost to the disease in 2020 [1–2]. While the average mortality rate hovers around 30%, this rate exhibits substantial variation contingent upon the economic status of the country [3–5]. In India, the recorded incidence and mortality of breast cancer in 2020 were 178,361 and 90,408, respectively [1–2]. These figures underscore a concerning statistic, revealing that roughly one in two women diagnosed with breast cancer succumb to the disease.
Early detection through community screening has been observed to be one of the crucial factors in reducing breast cancer mortality in developed countries [6–11]. In the United States, the mortality rate witnessed a decline from 33.7 per 100,000 women in 1989 to 25.5 per 100,000 women in 2020 [2, 12]. Similar downward trends in estimated annual percentage change (EAPC) were observed in other developed nations. However, such comprehensive data are often unavailable for many developing countries due to a lack of statistics, and where available, mortality rates often exhibit an upward trajectory.
The success of community screening in developed countries has been largely attributed to the implementation of mammography screening programs [6–11]. However, implementation of mammography in developing countries like India encounters several obstacles such as (a) high cost of mammography machines, (b) need for highly skilled expertise for both imaging and interpretation, (c) requirement for dedicated infrastructure with radiation shielding and high-voltage power, (d) lack of portability hindering rural screening camps, and (e) low screening uptakes due to physical discomfort associated with breast compression, fear of radiation-induced cancers, and absence of privacy during imaging leading to embarrassment. Moreover, mammography is generally not recommended and has been found to be less accurate in detecting breast cancers in young women or those with dense breast tissue [13–14]. However, an increasing number of young women are diagnosed with breast cancer in developing countries such as India, where approximately 46% of detected cancers occur in women under the age of 50 [15–16]. Furthermore, breast cancers detected in younger women tend to be more aggressive in nature [15].
Owing to the aforementioned challenges, clinical breast examination (CBE) by hand remains the recommended preliminary modality for breast cancer detection in India, as per the World Health Organization [16–17]. Although CBE is advantageous owing to its ease of implementation, its primary drawback is its low sensitivity (~ 50%) for detecting breast cancers [18–19]. In addition, CBE is subjective and lacks quantifiability, making standardization difficult. This subjectivity has led to low confidence among healthcare workers in conducting breast cancer screening. With an estimated 25% [20] increase in breast cancer incidence by 2040, there is a pressing need for alternative imaging systems that are better suited for countries like India.
In recent years, Thermalytix has emerged as an affordable and automated test for breast cancer detection in developing countries [21, 22]. Thermalytix leverages advances in thermal imaging hardware [23] and combines it with advancements in artificial intelligence (AI). It operates on the principle that the cancerous activities facilitated by the increased blood supply in the form of angiogenesis generate more heat, which traverses to the breast surface [23–24]. The latest thermal cameras are capable of capturing minute temperatures of up to 0.05 ℃ and the use of the sophisticated AI algorithms helps in understanding different heat patterns. Despite skepticism surrounding manual thermography [24], it is the combination of thermal imaging with AI, as exemplified by Thermalytix, which sets it apart.
Thermalytix has been validated in multiple prospective clinical and population-based screening studies. In a retrospective study conducted by Kakileti et al [25] on 470 women, encompassing both symptomatic and asymptomatic women, Thermalytix resulted in an overall sensitivity and specificity of 91.02% and 82.39%. When the results were computed only for 232 asymptomatic screening populations, the sensitivity was 100%, with a corresponding specificity of 92.41%. In a study [26] involving 13932 women participating in community-based screening camps, Thermalytix classified 625 (4.5%) women as suspicious, and recommended follow-up diagnostic tests. Of the 117 women for whom the reports were available, 74 benign and 4 malignant lesions were detected, resulting in a positive predictive value (PPV) of 66.7% in detecting benign and malignant breast lesions. In another community-based screening study [27] involving 6935 women, Thermalytix positivity rate and PPV were found to be 1.2% and 81.81%, respectively. Bansal et al. [28] evaluated the performance of Thermalytix on 459 women comprising both dense and fatty tissues and observed a sensitivity and specificity of 95.24% and 88.58%, respectively. Interestingly, in the 37 women where mammography was inconclusive or BIRADS 0, Thermalytix detected all three histopathologically confirmed malignancies with a sensitivity of 100%. In a prospective study [29] on 258 symptomatic women, Thermalytix demonstrated non-inferiority in sensitivity when compared to mammography. Finally, when the Thermalytix scores were utilized to compute a personalized breast cancer risk score on 769 women [30], it was found to be 20% better when compared to traditional age normalized risk score.
Furthermore, an independent health assessment study demonstrated that Thermalytix is more cost effective than Mammography at INR 8403.57 per DALY averted with a Net Monetary Benefit (NMB) of INR.1763.69 [31]. In a study by Davalagi et al. [32], the Strengths, Weaknesses, Opportunities, and Challenges (SWOC) analysis of Thermalytix involving 768 women identified reduced costs for screening services and the involvement of female self-help groups (SHGs) as significant strengths. While no specific weaknesses related to Thermalytix were identified, poor breast health awareness among women and stigma associated with breast cancer were acknowledged as potential weaknesses. These results, coupled with the inherent advantages of Thermalytix, such as low-cost, automated reporting, portability, non-invasiveness, privacy awareness, and radiation-free nature of the imaging position Thermalytix a suitable modality for low- and middle-income countries.
In this paper, we present a comprehensive analysis of screening outcomes for a substantial cohort of over 100,000 women using the Thermalytix test. This study is unique in multiple aspects, as follows:
-
It reflects real-world implementation of breast cancer screening using Thermalytix on a large-scale population.
-
It stands out as one of the very few large-scale studies leveraging artificial intelligence (AI) for the analysis of data from 100,000 participants.
-
The large-scale nature of the study allows for the assessment of its impact on breast cancer awareness among the screened population.
-
The data used for analysis are from 100,000 screenings conducted in the last five years, making this study one of the most recent and extensive studies conducted in India.
-
The data included a diverse range of participants, encompassing various demographics, socio-economic backgrounds, and geographical locations.
The insights derived from this unique dataset contribute significantly to our understanding of breast cancer detection and emphasize the potential impact of AI-based screening methodologies.