Invasive lobular carcinoma (ILC) is the second most common type of breast cancer, accounting for approximately 10%–15% of all breast cancer cases (1). Unlike its counterpart, invasive ductal carcinoma (IDC), ILC is often more challenging to detect and diagnose due to its unique biological characteristics and growth patterns. ILC tends to infiltrate breast tissue in a diffuse and linear fashion, lacking the distinct masses or lumps that are typically associated with IDC. This subtle and diffuse infiltration often results in delayed diagnosis, making it critical to explore the full spectrum of diagnostic tools available (2).

Over the years, various imaging and diagnostic techniques have been employed to improve the detection and characterization of lobular carcinoma. Traditional methods such as mammography and ultrasound remain first-line approaches; however, the limitations of these modalities, particularly in cases of ILC, necessitate the inclusion of more advanced and specialized diagnostic tools. Magnetic resonance imaging (MRI), digital breast tomosynthesis (DBT), and molecular imaging techniques have emerged as valuable adjuncts in the evaluation of ILC (3). Additionally, biopsy methods—guided by both imaging and histopathology—remain essential for definitive diagnosis (1, 2).

This paper aims to provide an overview of the current diagnostic modalities available for the detection and diagnosis of invasive lobular carcinoma. By exploring both conventional and emerging technologies, we hope to elucidate the strengths and limitations of each modality and propose an integrative approach to improve the early detection and management of ILC.

A direct comparison of the diagnostic capabilities of traditional and advanced imaging techniques shows that advanced modalities offer clear advantages. MRI provides the highest sensitivity among all modalities (33), especially for dense breast tissue, but its high cost and overdiagnosis potential remain concerns (17). DBT, though an improvement over mammography, is not specifically optimized for LC detection but has shown promise in detecting invasive lesions (22). CEM and BSGI both improve detection rates over traditional mammography, but BSGI stands out for its ability to detect lesions that are otherwise difficult to identify, making it an excellent tool for supplementary diagnostics. These findings are summarized in Table 1 .

The unique biological behavior and clinical characteristics of ILC present challenges in treatment planning and prognosis. However, AI and radiomics have shown promise in enhancing the diagnosis and risk stratification of ILC (40). While both radiomics and AI use imaging data to extract valuable insights, their approaches and applications differ. Radiomics involves the extraction of quantitative features from medical images, such as texture, shape, and intensity. These features provide detailed information about tumor characteristics that are not always visible to the human eye. Radiomics can enhance diagnostic accuracy, predict treatment responses, and help characterize tumors (36, 41). In contrast, AI integrates data from multiple sources—imaging, genomic, and clinical data—to perform predictive modeling and decision-making. AI uses these integrated datasets for tasks such as automated lesion detection, risk stratification, and prognosis, often incorporating radiomic features in its analysis. While radiomics focuses specifically on image-based features, AI combines multiple data types for broader clinical applications, including personalized treatment planning and long-term patient monitoring (42, 43).

Therefore, while radiomics provides detailed imaging data focused on tumor characteristics, AI goes further by combining imaging data with genomic and clinical information to make predictions and assist in clinical decision-making. AI can incorporate radiomic features into its analysis, but radiomics itself is focused on quantifying image-related tumor characteristics to inform diagnosis and prognosis.

Most AI-powered systems show promising results in improving diagnostic accuracy by analyzing large datasets from these imaging techniques, identifying subtle features in lobular carcinoma that may be missed by traditional methods (44, 45). A 2020 study by McKinney et al. evaluated an AI system for breast cancer screening across multiple international datasets. Their study revealed that the AI system outperformed human radiologists in identifying breast cancer, reducing false positives and false negatives, which is especially important for subtle and diffuse cancers like ILC (44). Similarly, Frazer et al. (45) evaluated deep learning-based AI techniques for breast cancer detection in mammograms and found that AI demonstrated high accuracy in breast cancer detection, with the potential to serve as a supplementary tool to radiologists in screening programs (45).

In addition to enhancing diagnostic accuracy, AI’s potential for improving diagnostic workflows in the context of LC includes aiding in risk stratification. For instance, AI models trained on imaging data from mammography, MRI, and ultrasound can analyze subtle features in lobular carcinoma, such as multifocality or bilaterality, which are crucial for accurate diagnosis and treatment planning (46, 47). By integrating AI into the diagnostic process, clinicians can make more informed decisions on the likelihood of disease recurrence and metastasis, thus refining risk stratification specifically for LC patients.

Moreover, AI enhances risk stratification by identifying patients at higher risk of recurrence or metastasis. Research suggests that ILC has a propensity for bilaterality and multifocality (43, 46). AI models can analyze imaging data to detect these features and help in more accurate risk assessment. For example, AI models trained on mammographic and MRI images can identify characteristics associated with a higher risk of contralateral breast cancer (47, 48).

Furthermore, AI can integrate clinical risk factors with molecular data to create comprehensive risk assessment tools. By incorporating a broader range of data, such as genetic mutations (e.g., CDH1 mutations) and hormonal factors, AI can provide more nuanced risk profiles (47, 48). This leads to more accurate predictions regarding disease progression and treatment outcomes.

AI’s ability to analyze histopathological data can also provide insights into tumor behavior and prognosis. For example, certain histological features, such as E-cadherin expression, are known to influence the aggressiveness of ILC (49). AI algorithms trained on histopathological images can identify these features and correlate them with clinical outcomes, enhancing prognostic accuracy (50).

In addition to AI, radiomics complements existing imaging by extracting quantitative features from medical images, offering deeper insights into tumor biology that traditional imaging may miss. It enhances diagnostic accuracy, improves risk stratification, and informs treatment decisions. For example, MRI-derived radiomic features can predict responses to neoadjuvant chemotherapy in lobular carcinoma patients (51), helping tailor personalized treatment plans.

Radiomics also plays a role in identifying breast cancer subtypes, aiding in the selection of therapeutic strategies (52), which is particularly useful for managing the subtle imaging characteristics of lobular carcinoma. Combined with AI, radiomics further enhances detection and characterization by analyzing vast datasets, offering improved sensitivity in dense breast tissue (53).

Moreover, radiomics helps monitor treatment responses and predict recurrence by tracking changes in tumor features, which is vital for managing lobular carcinoma’s varied growth patterns and therapy responses (54). When integrated with traditional imaging modalities like mammography and ultrasound, radiomics improves diagnostic workflows, enhancing the accuracy of interpretations and aiding in the distinction between benign and malignant lesions (55), leading to earlier detection and better outcomes.

LC is challenging to diagnose due to its subtle imaging characteristics, but advancements in imaging modalities hold promise for improving early detection and outcomes. While mammography and ultrasound remain important for initial assessments, advanced techniques like MRI, DBT, CEM, and BSGI offer higher diagnostic accuracy, particularly in dense breast tissue and high-risk patients. Integrating these methods, along with emerging technologies like AI and radiomics, into clinical workflows has significant potential to improve LC diagnosis. However, this integration requires careful planning and collaboration among clinicians, radiologists, and researchers to optimize diagnostic pathways and enhance patient outcomes.