RESUMEN

BACKGROUND: Spectral imaging of photon-counting detector CT (PCD-CT) scanners allows for generating virtual non-contrast (VNC) reconstruction. By analyzing 12 abdominal organs, we aimed to test the reliability of VNC reconstructions in preserving HU values compared to real unenhanced CT images. METHODS: Our study included 34 patients with pancreatic cystic neoplasm (PCN). The VNC reconstructions were generated from unenhanced, arterial, portal, and venous phase PCD-CT scans using the Liver-VNC algorithm. The observed 11 abdominal organs were segmented by the TotalSegmentator algorithm, the PCNs were segmented manually. Average densities were extracted from unenhanced scans (HUunenhanced), postcontrast (HUpostcontrast) scans, and VNC reconstructions (HUVNC). The error was calculated as HUerror=HUVNC-HUunenhanced. Pearson's or Spearman's correlation was used to assess the association. Reproducibility was evaluated by intraclass correlation coefficients (ICC). RESULTS: Significant differences between HUunenhanced and HUVNC[unenhanced] were found in vertebrae, paraspinal muscles, liver, and spleen. HUVNC[unenhanced] showed a strong correlation with HUunenhanced in all organs except spleen (r = 0.45) and kidneys (r = 0.78 and 0.73). In all postcontrast phases, the HUVNC had strong correlations with HUunenhanced in all organs except the spleen and kidneys. The HUerror had significant correlations with HUunenhanced in the muscles and vertebrae; and with HUpostcontrast in the spleen, vertebrae, and paraspinal muscles in all postcontrast phases. All organs had at least one postcontrast VNC reconstruction that showed good-to-excellent agreement with HUunenhanced during ICC analysis except the vertebrae (ICC: 0.17), paraspinal muscles (ICC: 0.64-0.79), spleen (ICC: 0.21-0.47), and kidneys (ICC: 0.10-0.31). CONCLUSIONS: VNC reconstructions are reliable in at least one postcontrast phase for most organs, but further improvement is needed before VNC can be utilized to examine the spleen, kidneys, and vertebrae.

Asunto(s)

RESUMEN

BACKGROUND: With photon-counting CT, spectral imaging is always available, and iodine maps with high spatial and spectral resolution can be generated. OBJECTIVES: The aim of this study was to investigate whether iodine uptake in different parenchymal patterns can be used to characterise parenchymal disease with increased lung attenuation. METHODS: 325 patients were scanned with a photon-counting CT using four scan protocols, all with lung parenchymal contrast. Lesions were classified into three basic patterns: consolidation, ground-glass opacities (GGO), and reticular pattern. Lesion classification was performed by 2 of 3 radiologists who were blinded to the diagnosis. Classification was performed twice using a 5-point Likert scale (with and without iodine maps). In case of disagreement, a third reader was consulted, and the decision was made by consensus. RESULTS: 206 lesions were found with a confirmed diagnosis (83 consolidations, 72 GGO, and 51 reticular). Diagnostic confidence improved when iodine maps were included in the evaluation. The mean Likert score increased significantly for all three basic patterns (consolidations: 3.3 vs. 3.9, GGO: 3.4 vs. 4.1, and reticular: 3.6 vs. 4.4, p < 0.001). However, the score for GGO and reticular pattern was downgraded in three and one cases, respectively. The downgrading occurred for morphologically uncertain GGO findings (3) and atelectasis (1) with inhomogeneous iodine uptake. In 29 lesions, the classification was changed when the iodine maps were included in the evaluation. CONCLUSION: Including iodine maps adds contrast uptake information and improves the diagnostic confidence of radiologists in the characterization of parenchymal pathologies. CLINICAL IMPACT: Iodine maps have the potential to provide complementary information for the interpretation of lung opacities with overlapping morphology.

RESUMEN

AIM: To assess the correlation of quantitative data of pulmonary Perfused Blood Volume (PBV) on Dual-Energy CT (DECT) datasets in patients with moderate - severe Pulmonary Emphysema (PE) with Lung Perfusion Scintigraphy (LPS) as the reference standard. The secondary endpoints are the correlation between the CT densitometric analysis and the visual assessment of parenchymal destruction with PBV. MATERIALS AND METHODS: Patients with moderate - severe PE candidate to Lung Volumetric Reduction (LVR), with available a pre-procedural LS and a contrast-enhanced DECT were retrospectively included. DECT studies were performed with a 3rd generation Dual-Source CT and the PBV was obtained with a 3-material decomposition algorithm. The CT densitometric analysis was performed with a dedicated commercial software (Pulmo3D). The Goddard Score was used for visual assessment. The perfusion LS were performed after the administration of albumin macroaggregates labeled with 99mTechnetium. The image revision was performed by two radiologists or nuclear medicine physicians blinded, respectively, to LS and DECT data. The statistical analysis was performed with nonparametric tests. RESULTS: Thirty-one patients (18 males, median age 69 y.o., interquartile range 62-71 y.o.) with moderate - severe PE (Median Goddard Score 14/20 and 31% of emphysematous parenchyma at quantitative CT) candidate to LVR were retrospectively included. The median enhancement on PBV was 17 HU. Significant correlation coefficients were demonstrated between lung PBV and LS, poor in apical regions (Rho = 0.1-0.2) and fair (Rho = 0.3-0.5) in middle and lower regions. No significant correlations were recorded between the CT densitometric analysis, the visual score, and the PBV. CONCLUSIONS: Lung perfusion with PBV on DECT is feasible in patients with moderate - severe PE candidate to LVR, and has a poor to fair agreement with LPS.

RESUMEN

Environment-sensitive probes are frequently used in spectral/multi-channel microscopy to study alterations in cell homeostasis. However, the few open-source packages available for processing of spectral images are limited in scope. Here, we present VISION, a stand-alone software based on Python for spectral analysis with improved applicability. In addition to classical intensity-based analysis, our software can batch-process multidimensional images with an advanced single-cell segmentation capability and apply user-defined mathematical operations on spectra to calculate biophysical and metabolic parameters of single cells. VISION allows for 3D and temporal mapping of properties such as membrane fluidity and mitochondrial potential. We demonstrate the broad applicability of VISION by applying it to study the effect of various drugs on cellular biophysical properties; the correlation between membrane fluidity and mitochondrial potential; protein distribution in cell-cell contacts; and properties of nanodomains in cell-derived vesicles. Together with the code, we provide a graphical user interface for facile adoption.

RESUMEN

Ventilation-perfusion SPECT with or without CT using technetium 99m (99mTc)-diethylenetriaminepentaacetic acid (DTPA) has been used to identify patterns typical of cardiopulmonary diseases, such as pulmonary embolism, pneumonia, heart failure, and obstructive lung disease. This case demonstrates the utility of a ventilation scan with SPECT/CT using 99mTc-DTPA for investigating the cause of a persistent complex pneumothorax in a patient with severe chronic obstructive pulmonary disease who recently underwent endobronchial valve placement. Keywords: CT-Spectral Imaging (Multienergy), SPECT/CT, Thorax, Lung Supplemental material is available for this article. © RSNA, 2024.

Asunto(s)

RESUMEN

The study of planets and small bodies within our Solar System is fundamental for understanding the formation and evolution of the Earth and other planets. Compositional and meteorological studies of the giant planets provide a foundation for understanding the nature of the most commonly observed exoplanets, while spectroscopic observations of the atmospheres of terrestrial planets, moons, and comets provide insights into the past and present-day habitability of planetary environments, and the availability of the chemical ingredients for life. While prior and existing (sub)millimeter observations have led to major advances in these areas, progress is hindered by limitations in the dynamic range, spatial and temporal coverage, as well as sensitivity of existing telescopes and interferometers. Here, we summarize some of the key planetary science use cases that factor into the design of the Atacama Large Aperture Submillimeter Telescope (AtLAST), a proposed 50-m class single dish facility: (1) to more fully characterize planetary wind fields and atmospheric thermal structures, (2) to measure the compositions of icy moon atmospheres and plumes, (3) to obtain detections of new, astrobiologically relevant gases and perform isotopic surveys of comets, and (4) to perform synergistic, temporally-resolved measurements in support of dedicated interplanetary space missions. The improved spatial coverage (several arcminutes), resolution (~ 1.2'' - 12''), bandwidth (several tens of GHz), dynamic range (~ 10 5) and sensitivity (~ 1 mK km s -1) required by these science cases would enable new insights into the chemistry and physics of planetary environments, the origins of prebiotic molecules and the habitability of planetary systems in general.

Our present understanding of what planets and comets are made of, and how their atmospheres move and change, has been greatly influenced by observations using existing and prior telescopes operating at wavelengths in the millimeter/submillimeter range (between the radio and infrared parts of the electromagnetic spectrum), yet major gaps exist in our knowledge of these diverse phenomena. Here, we describe the need for a new telescope capable of simultaneously observing features on very large and very small scales, and covering a very large spread of intrinsic brightness, in planets and comets. Such a telescope is required for mapping storms on giant planets, measuring the compositions of the atmospheres and plumes of icy moons, detecting new molecules in comets and planetary atmospheres, and to act as a complement for measurements by current and future interplanetary spacecraft missions. We discuss the limitations of currently-available millimeter/submillimeter telescopes, and summarize the requirements and applications of a new and larger, more sensitive facility operating at these wavelengths: the Atacama Large Aperture Submillimeter Telescope (AtLAST).

RESUMEN

Effective X-ray photon-counting spectral imaging (x-CSI) detector design involves the optimisation of a wide range of parameters both regarding the sensor (e.g., material, thickness and pixel pitch) and electronics (e.g., signal-processing chain and count-triggering scheme). Our previous publications have looked at the role of pixel pitch, sensor thickness and a range of additive charge sharing correction algorithms (CSCAs), and in this work, we compare additive and subtractive CSCAs to identify the advantages and disadvantages. These CSCAs differ in their approach to dealing with charge sharing: additive approaches attempt to reconstruct the original event, whilst subtractive approaches discard the shared events. Each approach was simulated on data from a wide range of x-CSI detector designs (pixel pitches 100-600 µm, sensor thickness 1.5 mm) and X-ray fluxes (106-109 photons mm-2 s-1), and their performance was characterised in terms of absolute detection efficiency (ADE), absolute photopeak efficiency (APE), relative coincidence counts (RCC) and binned spectral efficiency (BSE). Differences between the two approaches were explained mechanistically in terms of the CSCA's effect on both charge sharing and pule pileup. At low X-ray fluxes, the two approaches perform similarly, but at higher fluxes, they differ in complex ways. Generally, additive CSCAs perform better on absolute metrics (ADE and APE), and subtractive CSCAs perform better on relative metrics (RCC and BSE). Which approach to use will, thus, depend on the expected operating flux and whether dose efficiency or spectral efficiency is more important for the application in mind.

RESUMEN

As lithium-bearing minerals become critical raw materials for the field of energy storage and advanced technologies, the development of tools to accurately identify and differentiate these minerals is becoming essential for efficient resource exploration, mining, and processing. Conventional methods for identifying ore minerals often depend on the subjective observation skills of experts, which can lead to errors, or on expensive and time-consuming techniques such as Inductively Coupled Plasma Mass Spectrometry (ICP-MS) or Optical Emission Spectroscopy (ICP-OES). More recently, Raman Spectroscopy (RS) has emerged as a powerful tool for characterizing and identifying minerals due to its ability to provide detailed molecular information. This technique excels in scenarios where minerals have similar elemental content, such as petalite and spodumene, by offering distinct vibrational information that allows for clear differentiation between such minerals. Considering this case study and its particular relevance to the lithium-mining industry, this manuscript reports the development of an unsupervised methodology for lithium-mineral identification based on Raman Imaging. The deployed machine-learning solution provides accurate and interpretable results using the specific bands expected for each mineral. Furthermore, its robustness is tested with additional blind samples, providing insights into the unique spectral signatures and analytical features that enable reliable mineral identification.

RESUMEN

Spectral imaging has many applications, from methane detection using satellites to disease detection on crops. However, spectral cameras remain a costly solution ranging from 10 thousand to 100 thousand euros for the hardware alone. Here, we present a low-cost multispectral camera (LC-MSC) with 64 LEDs in eight different colors and a monochrome camera with a hardware cost of 340 euros. Our prototype reproduces spectra accurately when compared to a reference spectrometer to within the spectral width of the LEDs used and the ±1σ variation over the surface of ceramic reference tiles. The mean absolute difference in reflectance is an overestimate of 0.03 for the LC-MSC as compared to a spectrometer, due to the spectral shape of the tiles. In environmental light levels of 0.5 W m-2 (bright artificial indoor lighting) our approach shows an increase in noise, but still faithfully reproduces discrete reflectance spectra over 400 nm-1000 nm. Our approach is limited in its application by LED bandwidth and availability of specific LED wavelengths. However, unlike with conventional spectral cameras, the pixel pitch of the camera itself is not limited, providing higher image resolution than typical high-end multi- and hyperspectral cameras. For sample conditions where LED illumination bands provide suitable spectral information, our LC-MSC is an interesting low-cost alternative approach to spectral imaging.

RESUMEN

Fire blight is an infectious disease found in apple and pear orchards. While managing the disease is critical to maintaining orchard health, identifying symptoms early is a challenging task which requires trained expert personnel. This paper presents an inspection technique that targets individual symptoms via deep learning and density estimation. We evaluate the effects of including multi-spectral sensors in the model's pipeline. Results show that adding near infrared (NIR) channels can help improve prediction performance and that density estimation can detect possible symptoms when severity is in the mid-high range.

Asunto(s)

RESUMEN

BACKGROUND: Advances in computed tomography (CT) technology, particularly photon-counting CT (PCCT), are reshaping the possibilities for medical imaging. PCCT in spectral imaging enables the high-resolution visualization of tissues with material-specific accuracy. This study aims to establish a foundational approach for the in vivo visualization of intracranial blood using PCCT, focusing on non-enhanced imaging techniques and spectral imaging capabilities. METHODS: We employed photon-counting detector within a spectral CT framework to differentiate between venous and arterial intracranial blood. Our analysis included not only monoenergetic +67 keV reconstructions, but also images from virtual non-contrast and iodine phases, enabling detailed assessments of blood's characteristics without the use of contrast agents. RESULTS: Our findings demonstrate the ability of PCCT to provide clear and distinct visualizations of intracranial vascular structures. We quantified the signal-to-noise ratio across different imaging phases and found consistent enhancements in image clarity, particularly in the detection and differentiation of arterial and venous blood. CONCLUSION: PCCT offers a robust platform for the non-invasive and detailed visualization of intravascular intracranial blood. With its superior resolution and specific imaging capabilities, PCCT lays the groundwork for advancing clinical applications and research, notably in the diagnosis and management of intracranial disorders. This technology promises to improve diagnostic accuracy by enabling more precise imaging assessments.

RESUMEN

Liver fibrosis, a major global health issue, is marked by excessive collagen deposition that impairs liver function. Noninvasive methods for the direct visualization of collagen content are crucial for the early detection and monitoring of fibrosis progression. This study investigates the potential of spectral photoacoustic imaging (sPAI) to monitor collagen development in liver fibrosis. Utilizing a novel data-driven superpixel photoacoustic unmixing (SPAX) framework, we aimed to distinguish collagen presence and evaluate its correlation with fibrosis progression. We employed an established diethylnitrosamine (DEN) model in rats to study liver fibrosis over various time points. Our results revealed a significant correlation between increased collagen photoacoustic signal intensity and advanced fibrosis stages. Collagen abundance maps displayed dynamic changes throughout fibrosis progression. These findings underscore the potential of sPAI for the noninvasive monitoring of collagen dynamics and fibrosis severity assessment. This research advances the development of noninvasive diagnostic tools and personalized management strategies for liver fibrosis.

Asunto(s)

RESUMEN

Here we introduce a Raman spectroscopy approach combining multi-spectral imaging and a new fluorescence background subtraction technique to image individual Raman peaks in less than 5 seconds over a square field-of-view of 1-centimeter sides with 350 micrometers resolution. First, human data is presented supporting the feasibility of achieving cancer detection with high sensitivity and specificity - in brain, breast, lung, and ovarian/endometrium tissue - using no more than three biochemically interpretable biomarkers associated with the inelastic scattering signal from specific Raman peaks. Second, a proof-of-principle study in biological tissue is presented demonstrating the feasibility of detecting a single Raman band - here the CH2/CH3 deformation bands from proteins and lipids - using a conventional multi-spectral imaging system in combination with the new background removal method. This study paves the way for the development of a new Raman imaging technique that is rapid, label-free, and wide field.

RESUMEN

Refuse sorting is an important cornerstone of the recycling industry, but ever-changing refuse compositions and the desire to increase recycling rates still pose many unsolved challenges. The digitalisation of refuse sorting plants promises to overcome these challenges by optimising and automatically adapting the sorting process. This publication describes a system for image capturing, segmentation-based refuse recognition and data analysis of shredded refuse streams. The image capturing collects multispectral 2D and 3D images of the refuse streams on conveyor belts. The image recognition performs a semantic segmentation of the images to determine the refuse composition from the 2D images, whereas the 3D images approximate the volumes on the conveyor belts. The semantic segmentation is done by a combined convolutional neural network model, consisting of a foreground-background and a refuse class segmentation. Both models rely on synthetic training data to reduce the necessary amount of manually labelled training data, whereas the final segmentation performance reaches an Intersection over Union of up to 75%. The results of the semantic segmentation and volume estimation are combined with data of the shredding machinery by transforming it into a unified representation. This combined dataset is the basis for estimating the processed refuse masses from the semantic segmentation and volume estimation.

Asunto(s)

RESUMEN

OBJECTIVES: To assess the accuracy of a synthetic hematocrit derived from virtual non-contrast (VNC) and virtual non-iodine images (VNI) for myocardial extracellular volume (ECV) computation with photon-counting detector computed tomography (PCD-CT). MATERIALS AND METHODS: Consecutive patients undergoing PCD-CT including a coronary CT angiography (CCTA) and a late enhancement (LE) scan and having a blood hematocrit were retrospectively included. In the first 75 patients (derivation cohort), CCTA and LE scans were reconstructed as VNI at 60, 70, and 80 keV and as VNC with quantum iterative reconstruction (QIR) strengths 2, 3, and 4. Blood pool attenuation (BPmean) was correlated to blood hematocrit. In the next 50 patients (validation cohort), synthetic hematocrit was calculated using BPmean. Myocardial ECV was computed using the synthetic hematocrit and compared with the ECV using the blood hematocrit as a reference. RESULTS: In the derivation cohort (49 men, mean age 79 ± 8 years), a correlation between BPmean and blood hematocrit ranged from poor for VNI of CCTA at 80 keV, QIR2 (R2 = 0.12) to moderate for VNI of LE at 60 keV, QIR4; 70 keV, QIR3 and 4; and VNC of LE, QIR3 and 4 (all, R2 = 0.58). In the validation cohort (29 men, age 75 ± 14 years), synthetic hematocrit was calculated from VNC of the LE scan, QIR3. Median ECV was 26.9% (interquartile range (IQR), 25.5%, 28.8%) using the blood hematocrit and 26.8% (IQR, 25.4%, 29.7%) using synthetic hematocrit (VNC, QIR3; mean difference, -0.2%; limits of agreement, -2.4%, 2.0%; p = 0.33). CONCLUSION: Synthetic hematocrit calculated from VNC images enables an accurate computation of myocardial ECV with PCD-CT. CLINICAL RELEVANCE STATEMENT: Virtual non-contrast images from cardiac late enhancement scans with photon-counting detector CT allow the calculation of a synthetic hematocrit, which enables accurate computation of myocardial extracellular volume. KEY POINTS: Blood hematocrit is mandatory for conventional myocardial extracellular volume computation. Synthetic hematocrit can be calculated from virtual non-iodine and non-contrast photon-counting detector CT images. Synthetic hematocrit from virtual non-contrast images enables computation of the myocardial extracellular volume.

RESUMEN

In the forest industry, interspecific hybridization, such as Eucalyptus urograndis (Eucalyptus grandis × Eucalyptus urophylla) and Corymbia maculata × Corymbia torelliana, has led to the development of high-performing F1 generations. The successful breeding of these hybrids relies on verifying progenitor origins and confirming post-crossing, but conventional genotype identification methods are resource-intensive and result in seed destruction. As an alternative, multispectral imaging analysis has emerged as an efficient and non-destructive tool for seed phenotyping. This approach has demonstrated success in various crop seeds. However, identifying seed species in the context of forest seeds presents unique challenges due to their natural phenotypic variability and the striking resemblance between different species. This study evaluates the efficacy of spectral imaging analysis in distinguishing hybrid seeds of E. urograndis and C. maculata × C. torelliana from their progenitors. Four experiments were conducted: one for Corymbia spp. seeds, one for each Eucalyptus spp. batch separately, and one for pooled batches. Multispectral images were acquired at 19 wavelengths within the spectral range of 365-970 nm. Classification models based on Linear Discriminant Analysis (LDA), Random Forest (RF), and Support Vector Machine (SVM) was created using reflectance and reflectance features, combined with color, shape, and texture features, as well as nCDA transformed features. The LDA algorithm, combining all features, provided the highest accuracy, reaching 98.15% for Corymbia spp., and 92.75%, 85.38, and 86.00 for Eucalyptus batch one, two, and pooled batches, respectively. The study demonstrated the effectiveness of multispectral imaging in distinguishing hybrid seeds of Eucalyptus and Corymbia species. The seeds' spectral signature played a key role in this differentiation. This technology holds great potential for non-invasively classifying forest seeds in breeding programs.

Asunto(s)

RESUMEN

In this article, we demonstrate detection and identification of ten microplastic types directly in a water sample using an identification table derived from microplastic hyperspectral images. We selected a total of fourteen wavelengths which can be used to distinguish these ten microplastic types. We enhanced the visibility of these wavelengths by computationally removing water and baseline correcting with reflectance at 1550 nm. This method avoids, prevents, and eases most of the laborious sample preparation mandatory prior to analysis with robust techniques such as Raman spectroscopy and Fourier transform infrared spectroscopy. The ten different plastics were studied in water, first separately and then in a mixture. The microplastic concentrations varied depending on microplastic type and were kept <12 mg/ml per type. Finally, detection and identification were confirmed pixel-wise in a hyperspectral image of a realistic water matrix simulant including mixtures of only a few microplastic particles. All measurements have been performed with microplastics of different sizes and irregular shapes made in-house by milling commercial pellets and sheets. It enabled the establishment of a procedure for the identification of these vicious particles in real water samples. The present measurement setup of hyperspectral imaging and method of data analysis of a mixture of microplastics directly from a water-based sample may open a path towards fast, reliable, and on-site detection.

RESUMEN

Neuroinflammation is an underlying feature of neurodegenerative conditions, often appearing early in the aetiology of a disease. Microglial activation, a prominent initiator of neuroinflammation, can be induced through lipopolysaccharide (LPS) treatment resulting in expression of the inducible form of nitric oxide synthase (iNOS), which produces nitric oxide (NO). NO post-translationally modifies cysteine thiols through S-nitrosylation, which can alter function of the target protein. Furthermore, packaging of these NO-modified proteins into extracellular vesicles (EVs) allows for the exertion of NO signalling in distant locations, resulting in further propagation of the neuroinflammatory phenotype. Despite this, the NO-modified proteome of activated microglial EVs has not been investigated. This study aimed to identify the protein post-translational modifications NO signalling induces in neuroinflammation. EVs isolated from LPS-treated microglia underwent mass spectral surface imaging using time of flight-secondary ion mass spectrometry (ToF-SIMS), in addition to iodolabelling and comparative proteomic analysis to identify post-translation S-nitrosylation modifications. ToF-SIMS imaging successfully identified cysteine thiol side chains modified through NO signalling in the LPS treated microglial-derived EV proteins. In addition, the iodolabelling proteomic analysis revealed that the EVs from LPS-treated microglia carried S-nitrosylated proteins indicative of neuroinflammation. These included known NO-modified proteins and those associated with LPS-induced microglial activation that may play an essential role in neuroinflammatory communication. Together, these results show activated microglia can exert broad NO signalling changes through the selective packaging of EVs during neuroinflammation.

Asunto(s)

RESUMEN

Spectral imaging has revolutionisedvarious fields by capturing detailed spatial and spectral information. However, its high cost and complexity limit the acquisition of a large amount of data to generalise processes and methods, thus limiting widespread adoption. To overcome this issue, a body of the literature investigates how to reconstruct spectral information from RGB images, with recent methods reaching a fairly low error of reconstruction, as demonstrated in the recent literature. This article explores the modification of information in the case of RGB-to-spectral reconstruction beyond reconstruction metrics, with a focus on assessing the accuracy of the reconstruction process and its ability to replicate full spectral information. In addition to this, we conduct a colorimetric relighting analysis based on the reconstructed spectra. We investigate the information representation by principal component analysis and demonstrate that, while the reconstruction error of the state-of-the-art reconstruction method is low, the nature of the reconstructed information is different. While it appears that the use in colour imaging comes with very good performance to handle illumination, the distribution of information difference between the measured and estimated spectra suggests that caution should be exercised before generalising the use of this approach.

RESUMEN

Significance: Label-free multimodal imaging methods that can provide complementary structural and chemical information from the same sample are critical for comprehensive tissue analyses. These methods are specifically needed to study the complex tumor-microenvironment where fibrillar collagen's architectural changes are associated with cancer progression. To address this need, we present a multimodal computational imaging method where mid-infrared spectral imaging (MIRSI) is employed with second harmonic generation (SHG) microscopy to identify fibrillar collagen in biological tissues. Aim: To demonstrate a multimodal approach where a morphology-specific contrast mechanism guides a mid-infrared spectral imaging method to detect fibrillar collagen based on its chemical signatures. Approach: We trained a supervised machine learning (ML) model using SHG images as ground truth collagen labels to classify fibrillar collagen in biological tissues based on their mid-infrared hyperspectral images. Five human pancreatic tissue samples (sizes are in the order of millimeters) were imaged by both MIRSI and SHG microscopes. In total, 2.8 million MIRSI spectra were used to train a random forest (RF) model. The remaining 68 million spectra were used to validate the collagen images generated by the RF-MIRSI model in terms of collagen segmentation, orientation, and alignment. Results: Compared to the SHG ground truth, the generated MIRSI collagen images achieved a high average boundary F-score (0.8 at 4 pixels threshold) in the collagen distribution, high correlation (Pearson's R 0.82) in the collagen orientation, and similarly high correlation (Pearson's R 0.66) in the collagen alignment. Conclusions: We showed the potential of ML-aided label-free mid-infrared hyperspectral imaging for collagen fiber and tumor microenvironment analysis in tumor pathology samples.