Micro-computed Tomography in the Evaluation of Eosin-stained Axillary Lymph Node Biopsies of Females Diagnosed with Breast Cancer

Abstract

Histopathological investigation of metastasis in core needle axillary lymph node (ALN) biopsies is crucial for the prognosis and treatment planning of breast cancer patients. Biopsies are typically sliced and evaluated as two-dimensional (2D) images. Biopsy sampling errors and the limited view provided by 2D histology are leading factors contributing to false-negative results in the preoperative detection of metastatic lymph nodes and underestimation of metastatic foci.

In this proof-of-concept study, we aim to explore the technical feasibility and the potential capacities of tridimensional (3D) X-ray micro-computed tomography imaging to expedite error detection, enhancement of histopathological accuracy, and precise measurement of metastatic lesion on ALN core needle biopsies of two breast cancer patients. Our self-developed micro-CT protocol uses eosin for the first time, a common histological dye, to enhance 3D architecture of ALNs. Performed analysis on the images of the ALN biopsies involves cancer tissue segmentation, swift biopsy evaluation, and measurement of the metastatic longest diameter and deposit volume.

The eosin micro-CT protocol shows potential for an improved tumor deposit estimates, offering additional clinical value compared to standard 2D histology, however, further studies for validating this method are needed.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-79060-4.

Introduction

Breast cancer (BC) is the most common non-skin cancer worldwide, responsible for 6.9% of global cancer deaths in 2020¹. The prognosis for women with early-stage BC is strongly influenced by the presence and number of metastatic axillary lymph nodes (ALNs)². Histopathological analysis reveals metastatic ALNs in 24.8–35.5% of newly diagnosed BC cases^3–6. Preoperative assessment of the axilla is generally performed using ultrasound (US)². If suspicious ALNs are detected, a US-guided lymph node biopsy (US-LNB) is conducted for further histopathological evaluation². If no suspicious ALNs are detected, or if the US-LNB does not confirm metastasis, sentinel lymph nodes (S-LNB)—those first to drain from the tumor—are removed and biopsied during surgery for histopathological assessment.

Accurate nodal evaluation is critical for BC treatment planning: patients without micrometastases (metastatic focus longest diameter (LD) ≤ 2 mm) or macrometastases (LD > 2 mm) do not require adjuvant axillary radiotherapy, while those with confirmed micrometastases or 1–2 macrometastases receive radiotherapy. If more than two macrometastases are identified, axillary lymph node dissection (ALND) is performed². Detection of more than two macrometastases on US-LNB eliminates the need for intraoperative S-LNB².

A key challenge in current clinical practice is the risk of missing metastatic tissue during US-LNB or underestimating the extent of metastases during histopathological evaluation. These oversights can lead to nodal involvement false negatives (FN), which can cause delays or complications in treatment, increasing the risk to patients^7,8. The FN rate in nodal histopathology is approximately 30%, often resulting from sampling errors^9,10, thus, ongoing research is very oriented to enhance nodal status assessment with novel methodologies. Most representative examples include traditional radionics (US, MRI) or traditional histology powered with artificial intelligence to predict the nodal status. However, these methods frequently struggle to detect small metastatic foci or cannot capture the whole nodal architecture^11–15.

This unmet need motivates the development of standardized, clinically feasible imaging methods for nodal evaluation. Specifically, we believe that high-resolution 3D imaging techniques compatible with standard histopathology can offer a more comprehensive analysis in the preoperative assessment of ALN metastasis (ALNM), addressing the limitations of traditional 2D histopathology in assessing ALN biopsies, allowing direct observation of metastatic foci. In this proof-of-concept study, we explore the technical feasibility of a novel micro-Computed Tomography (micro-CT) imaging protocol to assess ALNM while compatible with the clinical standard. We focus on differentiating cancerous and healthy tissue components and measuring the LD and volume of metastatic deposits in paraffin-embedded ALN biopsy cores from BC patients. Our method is distinguished by the use of eosin as a contrast agent, making this the first study to utilize eosin-enhanced micro-CT for ALNM imaging.

Micro-CT is a well-established imaging technique that provides rapid, non-destructive 3D images of biological tissues at near-micron resolutions^16–18. In breast cancer (BC), its primary applications have been in intraoperative tumor margin assessment and the detection of mammary microcalcifications^19–23. However, studies demonstrating the utility of micro-CT for imaging nodal specimens are rare^24,25, yet critically important. These studies underscore the need for advanced micro-CT protocols that achieve cellular-level resolution while remaining compatible with clinical workflows. Although cellular-level resolutions are achievable^25–27, attaining such detail often demands extended imaging times or the use of contrast agents that may compromise the quality of subsequent histological analysis^28,29.

In this context, eosin-enhanced micro-CT stands as a strong candidate to be the method of choice. Eosin, a common histological dye, enables both 3D micro-CT imaging and 2D histology from a single application without causing significant detrimental effects on tissues^30,31. This approach, never before applied to nodal tissue and theoretically could achieve cellular-level resolution of metastatic foci. To our knowledge, such resolutions levels have not been reported with micro-CT for ALNs.

Confirming the technical feasibility and explore the analytical possibilities of eosin-enhanced micro-CT for ALNs is fundamental to justify further clinical research. Accuracy and sensitivivity validation studies of eosin micro-CT in detecting ALNMs can be determinant for optimizing imaging protocols, developing AI-based image analysis, and integrating multi-modal imaging in lymph node pathology in the future.

Materials and methods

Overview

Two experimental biopsies of metastasized axillary lymph nodes, from two different patients, were evaluated in this proof-of-concept study. The micro-CT protocol involved several steps. First, samples were stained with eosin. Second, they were embedded in paraffin, followed by 3D scanning. Lastly, histological processing was conducted. The 3D images were segmented (distinguishing cancer from healthy-appearing tissue), and analyzed using Avizo^® software (version 2022, Thermo Fisher Scientific) (Fig. 1).

Fig. 1.

Flowcharts of the micro-CT imaging protocol and the analysis of outcoming 3D images. The micro-CT protocol includes the staining of formalin-fixed axillary lymph node biopsies with eosin and micro-CT imaging. Quality control of the final image includes Signal-to-noise ratio (SNR) measurements. If the SNR threshold value (36) were not reached, the scanning should be repeated at an increased exposure time (1). If SNR criteria were not met afterwards, the sample should be re-stained for longer time and imaged again (2). The analysis of the final images consists of segmenting the cancerous tissue using gray values thresholding and manual adjustment of the segmentation mask.

Patients’ recruitment criteria

The current study was approved by the Regional Ethics Committee of Tampere University Hospital (study code R21030). Female patients diagnosed with breast cancer who were aged ≥ 18-years met the inclusion criteria if they had suspicious previously non-biopsied lymph nodes on MRI or contrast-enhanced spectral mammography (CESM) and were willing to participate in the study. If the patients did not meet the inclusion criteria, they were not approached. The patients were asked to give written informed consent to participate in the study. The study is concordant with the national laws and the Declaration of Helsinki. All personal data is handled in a manner that meets the EU General Data Protection Regulation 2016/679 (GDPR) on data protection.

Sample acquisition

Ultrasound-guided lymph node biopsy is performed during the preoperative assessment in concordance with the European guidelines³² with 16-gauge needles. Histopathological diagnoses are obtained from core needle biopsies prior to surgery and the final pathological confirmation of the lymph node status is determined using surgical samples. For the present proof-of-concept study, we obtained one additional axillary lymph node core needle biopsy from two patients with carcinomas of no special type (NST) who were subsequently diagnosed with axillary lymph node metastasis.

Micro-CT imaging protocol

Staining of biopsies with eosin

Contrast enhancement of axillary lymph node biopsies was achieved through the application of eosin stain. Initially, the biopsies were acidified using a 15 ml solution of 5% (v/v) glacial acetic acid in magnesium and calcium-free Phosphate Buffered Saline (PBS) for 48 h at 4 °C. Following the acidification step, the samples underwent a 1-hour rinse in PBS, followed by staining with a 30% (w/v) eosin solution (a mix of eosin Y and G [C.I. 45380], Certistain^®) in distilled water. The staining process was carried out at 4 °C with orbital rocking at 200 rpm for 48 h.

Mounting biopsies in the 3D scanner

After the application of eosin, the samples were embedded in paraffin³³ to maintain their positional conformation during the imaging. Samples were detached from the histological cassette and exceeding paraffin was trimmed off with a histological blade. Each biopsy was fitted into a 1 ml syringe and secured with a plunger (Fig. 2).

Fig. 2.

Micro-CT scanning setup. A) represents an axillary lymph node biopsy, stained with eosin and embedded in paraffin. The sample is inside a 1 ml syringe coupled onto a modular micro-CT mount. The paraffin-plunger compound plays a critical role in fixing the spatial conformation of the tissue during the scanning, preventing the sample to fold or move during the scanning. B) represents the micro-CT setup. A sample is inside the micro-CT scanner, placed between the X-ray source and detector.

Micro-CT imaging parameters

X-Radia microXCT-400 scanner was used for 3D image acquisitions. Mounted biopsies were accommodated inside the scanner enclosure 12 h before imaging to minimize potential micro-vibrational noise. 1601 projections (16-bit) were obtained for each tomographic scan. The electric potential and power were set at 40 kV/10 W, and the exposure time was fixed to 3 s per projection image. The source/detector distances were 40/8 mm, employing a 4x magnification lens. The micro-CT scanning time was 6 h per sample. The final pixel size for both images was 5.64.

Quality assessment of the micro-CT image

Quality control criteria was applied in the resulting 3D images. It was established accordingly with micro-CT’s manufacturer specifications and bibliography³⁴. An image was considered as acceptable when it reached a minimum amount of 5000 photon counts at the time of starting the imaging and when a minimum Signal-to-Noise-Ratio (SNR) value of 36 was obtained after the imaging. Avizo^® software (version 2022, Thermo Fisher Scientific) was used to estimate the SNR, which was computed as the ratio of the mean grayscale value of 5 representative tissue regions of interest (ROIs) to the averaged standard deviation of 5 random ROIs placed on the image’s background. All ROIs had a size of 50 × 50 × 50 pixels.

A contingency plan addressing potential image quality failure was also considered. The initial step would involve repeating the scan with an increased exposure time. If this adjustment proved insufficient to achieve the minimum Signal-to-Noise Ratio (SNR), next step would be to deparaffinize the sample. Subsequently, the sample would be re-stained with eosin for an additional 2 days before being re-imaged.

Re-orientation of 3D images to match the visualization plane with the histological slides

For each given sample, the micro-CT image was initially manually oriented to approximately align its visualization plane with the histological cutting plane. Subsequently, the 3D image was visualized throughout its tomographic stack to localize the tomographic section that displayed the closest land-marked morphological features with the histological slide (i.e., specific blood vessels). Orientation of the selected tomographic section was carefully refined by incrementally applying a 0.01° rotation step in the XY, XZ, and YZ planes until all the tissue structures and morphologies were distinctly present in both the histology and the corresponding tomographic section.

Histology

After the micro-CT imaging, the biopsy cores were re-embedded in paraffin for 2 h and 30 min at 62 °C to form a new block, then sliced at 4 μm thickness along their longest axis. Subsequent histological staining included classic hematoxylin and eosin (HE), as well as pan-cytokeratin immunohistochemistry, for a better visualization of the cancerous area (Anti PanCK clone BS5 Mouse monocle. BSH 7124 Nordic BioSite; 1:300 Ventana ab diluent; Ventana UltraView Universal DAB Detection kit). Final histological slides were digitized at 0.23 μm/pixel resolution, equivalent of 40X magnification (NanoZoomer S60, Hamamatsu) and saved in .tiff format for further analyses.

Analysis

Segmentation of the 3D micro-CT images

The segmentation aimed to distinguish between healthy-appearing and cancerous tissue. Prior to segmentation, each micro-CT image was median filtered. The segmentation method consisted of gray value band thresholding to select most of the cancerous tissue, followed by manual adjustment to include any missing cancerous regions in the selection. This method was first applied to an initial tomographic slice. Our pathologist validated the segmentation of that initial tomographic slice by comparing with the PanCK histological slide. Areas of the whole section and cancerous regions were measured in the initial tomographic slice and the histological slide and compared. The remaining tomographic slides were segmented sequentially, starting from the validated initial tomographic slide, using the thresholding and manual adjustment method.

Characterization of the grayscale values’ distribution within the micro-CT images

The quantitative analysis of the micro-CT images included the characterization of the gray values distribution of healthy and malignant tissue. Four regions were defined within the nodal tissue: (i) infiltrating cancerous tissue with high presence of intervening (desmoplastic/reactive) stroma; (ii) solid growth pattern of cancerous tissue with low presence of stroma; (iii) healthy-appearing (non-malignant) tissue with high presence of stroma and (iv) healthy-appearing tissue consisting mostly of lymphocytes with low presence of stroma. For each defined region, grayscale values were sampled from one ROI line of 300 μm. The distribution of these values along the lines where plotted and compared.

Comparison between 2D histological slides and 2D micro-CT analogue slices

Histological slides and their micro-CT analogue sections were compared side by side. The total area of the tissue and the diameter of arbitrary selected landmarks were measured on both image types and compared to detect possible effects of both imaging methods in the final size of the tissues.

3D measurements and morphological characterization of metastatic tissue

Measurements pertaining to the volume and surface area of the volume of the segmented cancerous and healthy-appearing tissue were acquired from the 3D image. Additionally, the analysis also focused on the visualization and the provision of a detailed morphological description of the cancerous pattern within the three-dimensional space.

Assessment of the longest diameter of the metastatic lesion (3D vs. 2D)

As the longest diameter of the metastatic lesion (LD) in the lymph node is of high prognostic value in breast cancer, it was measured in both the immunostained histological slide immunostained with PanCK and the 3D micro-CT images and compared. In the micro-CT images, the LD was measured as the straight line extending from the most apical point to the most basal point of the 3D cancer-segmented imaged. LD measurement in the histology was done accordingly to the clinical standard.

Results

Clinical features of the patients at the time of the study

The clinical features of the primary tumors and the axillary lymph nodes of the patients included in this study are disclosed in Table 1.

Table 1.

Clinical features of the patients at the time they were included in the study. ALND: axillary lymph node dissection; NST: Carcinoma of no special type; DCIS: Carcinoma in situ; ER: estrogen receptor; PR: progesterone receptor; HER2: human epidermal growth factor receptor 2; MKI67: cell proliferation marker of gene Ki-67.

	Patient 1	Patient 2
Metastasis at clinical biopsy of the lymph node	no	yes
Tumor type	NST + DCIS	NST
Tumor grade on biopsy	3	3
ER %	0	0
PR %	0	0
HER2	negative	negative
MKI67%	73	62
Breast operation	mastectomy	resection
Axillary operation	ALND	ALND

Quality of the micro-CT and histological imaging

Both specimens preserved their structural conformation during the micro-CT scanning and subsequent histology. The micro-CT-scanned samples 1 and 2 had SNR values of 94.54 and 79.11, respectively. The micro-CT images clearly displayed the main nodal architecture, and the histological images did not present major artifacts as confirmed by an experienced pathologist.

2D histological slides vs. their 2D micro-CT analogues

Direct comparison of the histological slides with their best-matching tomographic sections (Fig. 3) revealed a high degree of morphological resemblance. For sample 1, whole tissue area and metastatic lesion area of the immunoassayed histological slides were 9.52% (6.3 mm² vs. 5.7 mm²) and 12.5% larger than the micro-CT counterparts. For sample 2, the whole tissue area and metastatic lesion were respectively 20.75% (5.3 mm² vs. 4.2 mm²) and 18.88% (2.48 mm² vs. 2.02) larger in the histological slide (see Table 2).

Fig. 3.

Comparative analysis between the histology and eosin-enhanced micro-CT of the 2 metastatic axillary lymph node core biopsies included in the study. The graph in (A) shows the distribution of grayscale values plotted along 4 different lines of 300 μm. The lines collect the gray values of 4 regions of interest (ROI) on the micro-CT image (B). Each plotted line on the graph is also marked in the micro-CT image (B). The 4 ROIs are defined by the presence or apparent absence of cancer and the high or low presence of stroma. (C) is the histological slide analogue to the micro-CT section (B), stained by pan-Cytokeratin immunohistochemistry. (D), (F) and (H) are magnified regions of the micro-CT image (B) and their histological analogues, stained with hematoxylin and eosin, are displayed on (E), (G) and (I). Same arbitrary selected structures/landmarks have been marked and measured on both micro-CT and histological magnified images.

Table 2.

Comparison of histological and Micro-CT measurements for metastatic lesion and tissue parameters with percentage differences (Δ%). PanCK: Pan-cytokeratin reactive immunohistochemical staining; Δ% = ( (histological value - Micro-CT value) / histological value) × 100.

			Histological slide (PanCK)	Micro-CT analogue slice	Δ %
SAMPLES	1	Metastatic lesion area (mm 2 )	2.00	1.75	12.50
		Total tissue area (mm²)	6.30	5.70	9.52
		LD of the metastatic lesion (mm)	3.30	4.90	-48.48
		Total length of the tissue (mm)	8.60	8.50	1.16
	2	Metastatic lesion area (mm 2 )	2.49	2.02	18.88
		Total tissue area (mm²)	5.30	4.20	20.75
		LD of the metastatic lesion (mm)	4.90	5.60	-14.29
		Total length of the tissue (mm)	7.40	7.30	1.35

The diameters of the landmarked structures shared between the histological slides and their micro-CT analogues also presented dimensional differences (Fig. 3). In sample 1, all the measured structures were larger on histology. In sample 2, 1 out of the 3 measured landmarks presented a marked increase in the histological image. The overall differences between the histological slides and their micro-CT analogues were to some extend expected as histological slides tend to dilate and suffer inconsistent deformations while being processed (see discussion).

Characterization of the grayscale values distribution on the micro-CT images

The measured grayscale presented similar distribution patterns in both samples (Fig. 3). Healthy-appearing regions with low presence of stroma showed the highest micro-CT gray values. The gray values of the healthy-appearing and cancerous areas were overlapping in the presence of high stroma (Fig. 3).

Visualization, segmentation, longest diameter, and volume of the lymphoid metastatic tissue on micro-CT

Three-dimensional visualization of the micro-CT images enabled a rapid confirmation of lymphoid tissue in the biopsies and allowed the segmentation and evaluation of the healthy-appearing and cancerous tissues. The segmentation mask of the cancerous tissue was defined by gray values thresholding and manual adjustment guided by the histological images. Manual adjustment was needed because the threshold-based masking was partly insufficient due to the overlap of gray values in healthy-appearing and cancerous regions rich in stroma. The outcome of the segmentation revealed visually different growth patterns of cancer between both samples (Fig. 4). In sample 1, there was a discernable presence of solid growth, which was displayed as a three-dimensional, thick branching structure that channeled between healthy-appearing areas. In contrast, sample 2 had a predominant infiltrative-like pattern that expanded over most of the nodal tissue as confirmed by histology with hematoxylin and eosin (HE) and panCK-immunohistochemistry.

Fig. 4.

3D Micro-CT images of 2 metastatic axillary lymph node core biopsies and their 2D cross-sections. Panel (A) displays the three-dimensional image of the biopsy, along with four consecutive sections, which align with the cutting plane of the post-micro-CT histology. In panel (B), the micro-CT sections are perpendicular to those in (A). (C) and (D) panels are the analogues of (A) and (B) respectively and show the result of the segmentation (healthy-appearing vs. cancer). Notice how the morphological complexity and extension of cancer differ between both samples. The visualization of the nodal biopsies in 3D facilitates the early detection of sampling errors, the location and accurate measurement of critical histopathological markers (i.e. longest diameter of the metastatic lesion) and potentially enables a more realistic assessment of the metastatic extension.

The total volume of sample 1 was 607.2 mm³ of which 195.7 mm³ (32.2%) was deemed cancerous. The total volume of sample 2 was 574.0 mm³ of which 361.1 mm³ (62.9%) was categorized as cancerous tissue (Table 3). Final LD measurements differed between the immunoassayed histology and the micro-CT. For sample 1, the LD of the metastatic deposit measured on micro-CT (4.9 mm) was 48.48% larger than on histology (3.3 mm). For sample 2, the LD of the metastatic deposit in the micro-CT image (5.6 mm) was 14.29% larger than on histology (4.9 mm) (Table 2).

Table 3.

Volume and surface area measurements of the segmented healthy-appearing and cancer tissue of the axillary lymph node biopsies micro-CT images.

			Volume		Surface area
			mm³	%	mm²	%
SAMPLES	1	Healthy-appearing	411.5	67.8	969.8	51.6
		Cancer	195.7	32.2	908.5	48.4
		Total	607.2	100	1878.3	100.0
	2	Healthy-appearing	212.9	28.3	373.2	44.2
		Cancer	361.1	71.7	470.5	55.8
		Total	574.0	100	843.7	100.0

Discussion

In this proof-of-concept study, we present a pioneering eosin-based X-ray micro-CT method designed for the comprehensive three-dimensional (3D) visualization of distinct tissue compartments of axillary lymph nodes of female breast cancer patients with confirmed metastatic deposits. To the best of our knowledge, this is the first study showing the validity of X-ray micro-CT imaging in the 3D visualization of metastatic cancerous foci in the axillary lymph nodes using eosin as contrast enhancement stain. The good quality of the histopathological images obtained after micro-CT underscored the successful integration of our micro-CT protocol with subsequent histological analysis. In addition, the confirmed feasibility of analyzing metastatic deposits in the micro-CT images, strengths the potential of this approach to be used for preoperative ALN histopathology, encouraging further studies to validate the sensitivity and accuracy of this method.

There are several novel approaches for assessing lymph nodes in BC. The combination of advanced imaging techniques with artificial intelligence (AI) and radiomics has garnered significant attention for predicting lymph node status. Some studies explore ultrasound (US)-based methods to evaluate nodal status in clinically negative BC patients¹¹, while others focus on predicting the nodal involment using deep learning models on US images¹². Parallelly, advancements in MRI-based radiomics show promise as a non-invasive tool for predicting lymph node status in BC. Collectively, these studies underscore the growing role of AI and radiomics in enhancing the diagnostic and prognostic capabilities of breast cancer imaging. However, while useful for predicting nodal status, standard radiological approaches usually lack sensitivity to detect or visualize micrometastatic deposits. In this regard, micro-CT offers the advantage of providing cellular-level resolutions^{16–18,25–27}, potentially playing a crucial role in the enhanced evaluation and detection of micrometastatic foci.

Certainly, micro-CT has been applied to imaging lymph nodes, with two notable methodologies identified in the literature: (i) phase-contrast synchrotron micro-CT³⁵, conducted in Grenoble, France, and (ii) micro-CT using Lugol’s iodine (I₂-IK) as a contrast agent²⁴. While the first approach is promising, it is inherently not compatible with routine clinical use. The second approach demonstrated the critical role of contrast agents in effectively outlining metastatic foci. Both studies emphasized the need for further technical development and validation of micro-CT techniques capable of achieving cellular-level imaging.

Higher micro-CT resolutions are bounded to higher lens magnifications and longer imaging times. Preoperatively acquired ALN biopsies are smaller, and their analysis is less time-sensitive than intraoperative samples. For these reasons, we identified in preoperative ALN biopsies an opportunity to apply higher micro-CT imaging resolutions with practical clinical applications. To achieve this, the use of fully histopathology-compatible contrast agents is fundamental, and eosin stands as a promising candidate^30,31.

The use of standard iodine-based contrast agents comes with limitations. Iodine stains are known to cause tissue shrinkage²⁸. furthermore, for optimal micro-CT subsequent histology, iodine must be decolored, which extends the overall sample processing time²⁹. In contrast, as previous research has shown^30,31, we also demonstrated that standard histological processing after eosin-based micro-CT imaging can be performed without complex laboratory protocols.

While using X-ray micro-CT for 3D biopsy morphometrics is not new^36,37, we applied this technique specifically to measure the largest diameters (LDs) of foci in ALN metastases. The LDs measurement in 3D images, to our knowledge, has never been reported. Measured LDs in the micro-CT image were larger than those observed in histological slides. This discrepancy likely arises from the fact that 3D imaging allows for the visualization of the entire biopsy, offering a theoretically more realistic measurement of LDs than the 2D histological slices. However, validating the accuracy of this type of analysis requires of further validation studies.

In contrast to the LD measurements, the area values of the histological slides were larger than those measured in the micro-CT images. Extensive literature exists on the effects of standard histology on paraffin-embedded (PE) samples, which tend to expand and undergo inconsistent morphological alterations when sliced and mounted^38–41. Although thorough investigation of the impact of these alterations on the area measurements was not object of this study, the larger tissue areas observed in the histology are consistent with this phenomenon. Additionally, the differences in metastatic tissue areas could also be attributed to segmentation accuracy and sensitivity of the 3D imaging, which warrants further validation of the eosin-micro-CT method in assessing ALN metastases.

Despite these challenges, we believe that the eosin-micro-CT approach holds great potential for a more consistent assessment of the LDs of cancerous deposits, as it avoids the tissue distortions and the limited visualization inherent to 2D histology.

From a practical perspective, embedding samples in paraffin after eosin staining, prior to micro-CT imaging, washighly convinient. This process stabilized the tissue’s physical conformation during both the micro-CT scan and subsequent histological analysis, facilitating the micro-CT and histological planes alignment. Moreover, in the literature is confirmed that paraffin can enhance micro-CT imaging quality due to its mass attenuation coefficient, which differs significantly from that of soft tissues⁴². Since paraffin embedding is a routine practice in axillary lymph node biopsies, this approach further facilitates the integration of micro-CT with standard ALN analysis.

Our study opens new possibilities for histopathological imaging with potential clinical implications, pending further validation. The eosin-micro-CT method, as a 3D imaging technique, may offer a more precise histopathological assessment of axillary lymph node metastases.

Limitations of the study

Our proposed eosin-based X-ray micro-CT method has shown technical compatibility with standard clinical histopathology. However, further validation is required to confirm the method’s accuracy and sensitivity in detecting metastatic foci. A limitation of the current study is the small sample size, given its proof-of-concept design. Future research should include larger cohorts encompassing both metastatic and non-metastatic deposits to provide robust clinical validation. Additionally, larger sample sizes are crucial for assessing the prognostic value and potential predictive role of 3D measurements.

While the micro-CT images in this study achieved a resolution of approximately 6 microns—sufficient for theoretical observation of some cellular structures—this resolution remains a limitation. Enhancing the clinical applicability of this technique will require higher resolution and further validation.

The segmentation method used in this study is not yet optimal for clinical practice. The overlapping gray values of stroma-rich cancerous and healthy-appearing areas are likely due to the use of eosin, a non-specific stain with an affinity for the negatively charged amino groups of cytoplasmic proteins³⁰. As cell density decreases, the eosin concentration diminishes, causing both infiltrative cancer and some healthy-appearing, stroma-rich regions to exhibit similar staining patterns under the micro-CT. This underscores the need for further investigation into more specific contrast agents or advanced segmentation techniques to better delineate metastatic foci.

Looking ahead, addressing the labor-intensive process of image orientation and segmentation in micro-CT will require the development and validation of automated co-registration and segmentation tools. These advancements are essential for translating this method into practical clinical use.

Conclusion

The present study confirms the technical feasibility of our novel eosin-staining-based 3D micro-CT imaging for visualizing metastatic foci in preoperative ALN biopsies and its compatibility with standard histology. This approach has potential to facilitate more accurate estimates of the largest diameter of the metastatic foci, and volumetric measurements of ALN tumor deposits, which may have additional clinical value compared to standard 2D histology. Future research should focus on validating the accuracy and sensitivity of this technique, as well as developing faster segmentation tools for identifying metastatic foci.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

SL-C: Conceptualization, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing.AS: Conceptualization, Investigation, Data curation.OA: Conceptualization, Supervision, Writing – original draft, Writing – review & editing.MH: Conceptualization, Investigation, Writing – review & editing.IRK: Project administration, Resources, Supervision, Writing – review & editing.JH: Conceptualization, Resources, Supervision, Writing – review & editing.TT: Project administration, Supervision, Writing – review & editing.

Data availability

The data that support the findings of this study have been provided in this submission, without restrictions. The raw data, processed data, and any supplementary materials required to reproduce the reported results are accessible for scrutiny and validation. Researchers interested in accessing the data for non-commercial purposes may contact the corresponding author for further details.We encourage fellow researchers to utilize and build upon the data provided here, promoting transparency and facilitating the advancement of scientific knowledge in the respective field.We are committed to facilitating data access for the purpose of scientific validation and reproducibility, while also respecting any constraints that necessitate restricted access.

Declarations

Competing interests

The authors declare no competing interests.