Toward noncontact macroscopic imaging of multiple cancers using multi-spectral inelastic scattering detection.

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.

Macroscopic inelastic scattering imaging using a hyperspectral line-scanning system identifies invasive breast cancer in lumpectomy and mastectomy specimens.

Significance: Of patients with early-stage breast cancer, 60% to 75% undergo breast-conserving surgery. Of those, 20% or more need a second surgery because of an incomplete tumor resection only discovered days after surgery. An intraoperative imaging technology allowing cancer detection on the margins of breast specimens could reduce re-excision procedure rates and improve patient survival. Aim: We aimed to develop an experimental protocol using hyperspectral line-scanning Raman spectroscopy to image fresh breast specimens from cancer patients. Our objective was to determine whether macroscopic specimen images could be produced to distinguish invasive breast cancer from normal tissue structures. Approach: A hyperspectral inelastic scattering imaging instrument was used to interrogate eight specimens from six patients undergoing breast cancer surgery. Machine learning models trained with a different system to distinguish cancer from normal breast structures were used to produce tissue maps with a field-of-view of 1 cm 2 classifying each pixel as either cancer, adipose, or other normal tissues. The predictive model results were compared with spatially correlated histology maps of the specimens. Results: A total of eight specimens from six patients were imaged. Four of the hyperspectral images were associated with specimens containing cancer cells that were correctly identified by the new ex vivo pathology technique. The images associated with the remaining four specimens had no histologically detectable cancer cells, and this was also correctly predicted by the instrument. Conclusions: We showed the potential of hyperspectral Raman imaging as an intraoperative breast cancer margin assessment technique that could help surgeons improve cosmesis and reduce the number of repeat procedures in breast cancer surgery.

In situ brain tumor detection using a Raman spectroscopy system-results of a multicenter study.

Safe and effective brain tumor surgery aims to remove tumor tissue, not non-tumoral brain. This is a challenge since tumor cells are often not visually distinguishable from peritumoral brain during surgery. To address this, we conducted a multicenter study testing whether the Sentry System could distinguish the three most common types of brain tumors from brain tissue in a label-free manner. The Sentry System is a new real time, in situ brain tumor detection device that merges Raman spectroscopy with machine learning tissue classifiers. Nine hundred and seventy-six in situ spectroscopy measurements and colocalized tissue specimens were acquired from 67 patients undergoing surgery for glioblastoma, brain metastases, or meningioma to assess tumor classification. The device achieved diagnostic accuracies of 91% for glioblastoma, 97% for brain metastases, and 96% for meningiomas. These data show that the Sentry System discriminated tumor containing tissue from non-tumoral brain in real time and prior to resection.

Robot-assisted biopsy sampling for online Raman spectroscopy cancer confirmation in the operating room.

PURPOSE: Cancer confirmation in the operating room (OR) is crucial to improve local control in cancer therapies. Histopathological analysis remains the gold standard, but there is a lack of real-time in situ cancer confirmation to support margin confirmation or remnant tissue. Raman spectroscopy (RS), as a label-free optical technique, has proven its power in cancer detection and, when integrated into a robotic assistance system, can positively impact the efficiency of procedures and the quality of life of patients, avoiding potential recurrence. METHODS: A workflow is proposed where a 6-DOF robotic system (optical camera + MECA500 robotic arm) assists the characterization of fresh tissue samples using RS. Three calibration methods are compared for the robot, and the temporal efficiency is compared with standard hand-held analysis. For healthy/cancerous tissue discrimination, a 1D-convolutional neural network is proposed and tested on three ex vivo datasets (brain, breast, and prostate) containing processed RS and histopathology ground truth. RESULTS: The robot achieves a minimum error of 0.20 mm (0.12) on a set of 30 test landmarks and demonstrates significant time reduction in 4 of the 5 proposed tasks. The proposed classification model can identify brain, breast, and prostate cancer with an accuracy of 0.83 (0.02), 0.93 (0.01), and 0.71 (0.01), respectively. CONCLUSION: Automated RS analysis with deep learning demonstrates promising classification performance compared to commonly used support vector machines. Robotic assistance in tissue characterization can contribute to highly accurate, rapid, and robust biopsy analysis in the OR. These two elements are an important step toward real-time cancer confirmation using RS and OR integration.

Spectral effects and enhancement quantification in healthy human saliva with surface-enhanced Raman spectroscopy using silver nanopillar substrates.

OBJECTIVES: Raman spectroscopy as a diagnostic tool for biofluid applications is limited by low inelastic scattering contributions compared to the fluorescence background from biomolecules. Surface-enhanced Raman spectroscopy (SERS) can increase Raman scattering signals, thereby offering the potential to reduce imaging times. We aimed to evaluate the enhancement related to the plasmonic effect and quantify the improvements in terms of spectral quality associated with SERS measurements in human saliva. METHODS: Dried human saliva was characterized using spontaneous Raman spectroscopy and SERS. A fabrication protocol was implemented leading to the production of silver (Ag) nanopillar substrates by glancing angle deposition. Two different imaging systems were used to interrogate saliva from 161 healthy donors: a custom single-point macroscopic system and a Raman micro-spectroscopy instrument. Quantitative metrics were established to compare spontaneous RS and SERS measurements: the Raman spectroscopy quality factor (QF), the photonic count rate (PR), the signal-to-background ratio (SBR). RESULTS: SERS measurements acquired with an excitation energy four times smaller than with spontaneous RS resulted in improved QF, PR values an order of magnitude larger and a SBR twice as large. The SERS enhancement reached 100×, depending on which Raman bands were considered. CONCLUSIONS: Single-point measurement of dried saliva with silver nanopillars substrates led to reproducible SERS measurements, paving the way to real-time tools of diagnosis in human biofluids.

Subsecond lung cancer detection within a heterogeneous background of normal and benign tissue using single-point Raman spectroscopy.

Significance: Lung cancer is the most frequently diagnosed cancer overall and the deadliest cancer in North America. Early diagnosis through current bronchoscopy techniques is limited by poor diagnostic yield and low specificity, especially for lesions located in peripheral pulmonary locations. Even with the emergence of robotic-assisted platforms, bronchoscopy diagnostic yields remain below 80%. Aim: The aim of this study was to determine whether in situ single-point fingerprint (800 to 1700 cm-1) Raman spectroscopy coupled with machine learning could detect lung cancer within an otherwise heterogenous background composed of normal tissue and tissue associated with benign conditions, including emphysema and bronchiolitis. Approach: A Raman spectroscopy probe was used to measure the spectral fingerprint of normal, benign, and cancer lung tissue in 10 patients. Each interrogated specimen was characterized by histology to determine cancer type, i.e., small cell carcinoma or non-small cell carcinoma (adenocarcinoma and squamous cell carcinoma). Biomolecular information was extracted from the fingerprint spectra to identify biomolecular features that can be used for cancer detection. Results: Supervised machine learning models were trained using leave-one-patient-out cross-validation, showing lung cancer could be detected with a sensitivity of 94% and a specificity of 80%. Conclusions: This proof of concept demonstrates fingerprint Raman spectroscopy is a promising tool for the detection of lung cancer during diagnostic procedures and can capture biomolecular changes associated with the presence of cancer among a complex heterogeneous background within less than 1 s.

Detection of SARS-CoV-2 in saliva by a low-cost LSPR-based sensor.

The SARS-CoV-2 pandemic started more than 3 years ago, but the containment of the spread is still a challenge. Screening is imperative for informed decision making by government authorities to contain the spread of the virus locally. The access to screening tests is disproportional, due to the lack of access to reagents, equipment, finances or because of supply chain disruptions. Low and middle-income countries have especially suffered with the lack of these resources. Here, we propose a low cost and easily constructed biosensor device based on localized surface plasmon resonance, or LSPR, for the screening of SARS-CoV-2. The biosensor device, dubbed "sensor" for simplicity, was constructed in two modalities: (1) viral detection in saliva and (2) antibody against COVID in saliva. Saliva collected from 18 patients were tested in triplicates. Both sensors successfully classified all COVID positive patients (among hospitalized and non-hospitalized). From the COVID negative patients 7/8 patients were correctly classified. For both sensors, sensitivity was determined as 100% (95% CI 79.5-100) and specificity as 87.5% (95% CI 80.5-100). The reagents and equipment used for the construction and deployment of this sensor are ubiquitous and low-cost. This sensor technology can then add to the potential solution for challenges related to screening tests in underserved communities.

Spectral focusing-based stimulated Raman scattering microscopy using compact glass blocks for adjustable dispersion.

Spectral focusing is a well-established technique for increasing spectral resolution in coherent Raman scattering microscopy. However, current methods for tuning optical chirp in setups using spectral focusing, such as glass rods, gratings, and prisms, are very cumbersome, time-consuming to use, and difficult to align, all of which limit more widespread use of the spectral focusing technique. Here, we report a stimulated Raman scattering (SRS) configuration which can rapidly tune optical chirp by utilizing compact adjustable-dispersion TIH53 glass blocks. By varying the height of the blocks, the number of bounces in the blocks and therefore path length of the pulses through the glass can be quickly modulated, allowing for a convenient method of adjusting chirp with almost no necessary realignment. To demonstrate the flexibility of this configuration, we characterize our system's signal-to-noise ratio and spectral resolution at different chirp values and perform imaging in both the carbon-hydrogen stretching region (MCF-7 cells) and fingerprint region (prostate cores). Our findings show that adjustable-dispersion glass blocks allow the user to effortlessly modify their optical system to suit their imaging requirements. These blocks can be used to significantly simplify and miniaturize experimental configurations utilizing spectral focusing.

Preclinical evaluation of Raman spectroscopy for pedicular screw insertion surgical guidance in a porcine spine model.

Significance: Orthopedic surgery is frequently performed but currently lacks consensus and availability of ideal guidance methods, resulting in high variability of outcomes. Misdirected insertion of surgical instruments can lead to weak anchorage and unreliable fixation along with risk to critical structures including the spinal cord. Current methods for surgical guidance using conventional medical imaging are indirect and time-consuming with unclear advantages. Aim: The purpose of this study was to investigate the potential of intraoperative in situ near-infrared Raman spectroscopy (RS) combined with machine learning in guiding pedicular screw insertion in the spine. Approach: A portable system equipped with a hand-held RS probe was used to make fingerprint measurements on freshly excised porcine vertebrae, identifying six tissue types: bone, spinal cord, fat, cartilage, ligament, and muscle. Supervised machine learning techniques were used to train-and test on independent hold-out data subsets-a six-class model as well as two-class models engineered to distinguish bone from soft tissue. The two-class models were further tested using in vivo spectral fingerprint measurements made during intra-pedicular drilling in a porcine spine model. Results: The five-class model achieved >96% accuracy in distinguish all six tissue classes when applied onto a hold-out testing data subset. The binary classifier detecting bone versus soft tissue (all soft tissue or spinal cord only) yielded 100% accuracy. When applied onto in vivo measurements performed during interpedicular drilling, the soft tissue detection models correctly detected all spinal canal breaches. Conclusions: We provide a foundation for RS in the orthopedic surgical guidance field. It shows that RS combined with machine learning is a rapid and accurate modality capable of discriminating tissues that are typically encountered in orthopedic procedures, including pedicle screw placement. Future development of integrated RS probes and surgical instruments promises better guidance options for the orthopedic surgeon and better patient outcomes.

In situ Raman spectroscopy and machine learning unveil biomolecular alterations in invasive breast cancer.

Significance: As many as 60% of patients with early stage breast cancer undergo breast-conserving surgery. Of those, 20% to 35% need a second surgery because of incomplete resection of the lesions. A technology allowing in situ detection of cancer could reduce re-excision procedure rates and improve patient survival. Aim: Raman spectroscopy was used to measure the spectral fingerprint of normal breast and cancer tissue ex-vivo. The aim was to build a machine learning model and to identify the biomolecular bands that allow one to detect invasive breast cancer. Approach: The system was used to interrogate specimens from 20 patients undergoing lumpectomy, mastectomy, or breast reduction surgery. This resulted in 238 ex-vivo measurements spatially registered with standard histology classifying tissue as cancer, normal, or fat. A technique based on support vector machines led to the development of predictive models, and their performance was quantified using a receiver-operating-characteristic analysis. Results: Raman spectroscopy combined with machine learning detected normal breast from ductal or lobular invasive cancer with a sensitivity of 93% and a specificity of 95%. This was achieved using a model based on only two spectral bands, including the peaks associated with C-C stretching of proteins around 940 cm - 1 and the symmetric ring breathing at 1004 cm - 1 associated with phenylalanine. Conclusions: Detection of cancer on the margins of surgically resected breast specimen is feasible with Raman spectroscopy.

A clinical Raman spectroscopy imaging system and safety requirements for in situ intraoperative tissue characterization.

Raman spectroscopy imaging is a technique that can be adapted for intraoperative tissue characterization to be used for surgical guidance. Here we present a macroscopic line scanning Raman imaging system that has been modified to ensure suitability for intraoperative use. The imaging system has a field of view of 1 × 1 cm2 and acquires Raman fingerprint images of 40 × 42 pixels, typically in less than 5 minutes. The system is mounted on a mobile cart, it is equiped with a passive support arm and possesses a removable and sterilizable probe muzzle. The results of a proof of concept study are presented in porcine adipose and muscle tissue. Supervised machine learning models (support vector machines and random forests) were trained and they were tested on a holdout dataset consisting of 7 Raman images (10 080 spectra) acquired in different animal tissues. This led to a detection accuracy >96% and prediction confidence maps providing a quantitative detection assessment for tissue border visualization. Further testing was accomplished on a dataset acquired with the imaging probe's contact muzzle and tailored classification models showed robust classifications capabilities with specificity, sensitivity and accuracy all surpassing 95% with a support vector machine classifier. Finally, laser safety, biosafety and sterilization of the system was assest. The safety assessment showed that the system's laser can be operated safetly according to the American National Standards Institute's standard for maximum permissible exposures for eyes and skin. It was further shown that during tissue interrogation, the temperature-history in cumulative equivalent minutes at 43 °C (CEM43 °C) never exceeded a safe threshold of 5 min.

Open-sourced Raman spectroscopy data processing package implementing a baseline removal algorithm validated from multiple datasets acquired in human tissue and biofluids.

Significance: Standardized data processing approaches are required in the field of bio-Raman spectroscopy to ensure information associated with spectral data acquired by different research groups, and with different systems, can be compared on an equal footing. Aim: An open-sourced data processing software package was developed, implementing algorithms associated with all steps required to isolate the inelastic scattering component from signals acquired using Raman spectroscopy devices. The package includes a novel morphological baseline removal technique (BubbleFill) that provides increased adaptability to complex baseline shapes compared to current gold standard techniques. Also incorporated in the package is a versatile tool simulating spectroscopic data with varying levels of Raman signal-to-background ratios, baselines with different morphologies, and varying levels of stochastic noise. Results: Application of the BubbleFill technique to simulated data demonstrated superior baseline removal performance compared to standard algorithms, including iModPoly and MorphBR. The data processing workflow of the open-sourced package was validated in four independent in-human datasets, demonstrating it leads to inter-systems data compatibility. Conclusions: A new open-sourced spectroscopic data pre-processing package was validated on simulated and real-world in-human data and is now available to researchers and clinicians for the development of new clinical applications using Raman spectroscopy.

Image-guided Raman spectroscopy navigation system to improve transperineal prostate cancer detection. Part 1: Raman spectroscopy fiber-optics system and in situ tissue characterization.

SIGNIFICANCE: The diagnosis of prostate cancer (PCa) and focal treatment by brachytherapy are limited by the lack of precise intraoperative information to target tumors during biopsy collection and radiation seed placement. Image-guidance techniques could improve the safety and diagnostic yield of biopsy collection as well as increase the efficacy of radiotherapy. AIM: To estimate the accuracy of PCa detection using in situ Raman spectroscopy (RS) in a pilot in-human clinical study and assess biochemical differences between in vivo and ex vivo measurements. APPROACH: A new miniature RS fiber-optics system equipped with an electromagnetic (EM) tracker was guided by trans-rectal ultrasound-guided imaging, fused with preoperative magnetic resonance imaging to acquire 49 spectra in situ (in vivo) from 18 PCa patients. In addition, 179 spectra were acquired ex vivo in fresh prostate samples from 14 patients who underwent radical prostatectomy. Two machine-learning models were trained to discriminate cancer from normal prostate tissue from both in situ and ex vivo datasets. RESULTS: A support vector machine (SVM) model was trained on the in situ dataset and its performance was evaluated using leave-one-patient-out cross validation from 28 normal prostate measurements and 21 in-tumor measurements. The model performed at 86% sensitivity and 72% specificity. Similarly, an SVM model was trained with the ex vivo dataset from 152 normal prostate measurements and 27 tumor measurements showing reduced cancer detection performance mostly attributable to spatial registration inaccuracies between probe measurements and histology assessment. A qualitative comparison between in situ and ex vivo measurements demonstrated a one-to-one correspondence and similar ratios between the main Raman bands (e.g., amide I-II bands, phenylalanine). CONCLUSIONS: PCa detection can be achieved using RS and machine learning models for image-guidance applications using in situ measurements during prostate biopsy procedures.

Image-guided Raman spectroscopy navigation system to improve transperineal prostate cancer detection. Part 2: in-vivo tumor-targeting using a classification model combining spectral and MRI-radiomics features.

SIGNIFICANCE: The diagnosis and treatment of prostate cancer (PCa) are limited by a lack of intraoperative information to accurately target tumors with needles for biopsy and brachytherapy. An innovative image-guidance technique using optical devices could improve the diagnostic yield of biopsy and efficacy of radiotherapy. AIM: To evaluate the performance of multimodal PCa detection using biomolecular features from in-situ Raman spectroscopy (RS) combined with image-based (radiomics) features from multiparametric magnetic resonance images (mpMRI). APPROACH: In a prospective pilot clinical study, 18 patients were recruited and underwent high-dose-rate brachytherapy. Multimodality image fusion (preoperative mpMRI with intraoperative transrectal ultrasound) combined with electromagnetic tracking was used to navigate an RS needle in the prostate prior to brachytherapy. This resulting dataset consisted of Raman spectra and co-located radiomics features from mpMRI. Feature selection was performed with the constraint that no more than 10 features were retained overall from a combination of inelastic scattering spectra and radiomics. These features were used to train support vector machine classifiers for PCa detection based on leave-one-patient-out cross-validation. RESULTS: RS along with biopsy samples were acquired from 47 sites along the insertion trajectory of the fiber-optics needle: 26 were confirmed as benign or grade group = 1, and 21 as grade group >1, according to histopathological reports. The combination of the fingerprint region of the RS and radiomics showed an accuracy of 83% (sensitivity = 81 % and a specificity = 85 % ), outperforming by more than 9% models trained with either spectroscopic or mpMRI data alone. An optimal number of features was identified between 6 and 8 features, which have good potential for discriminating grade group ≥1 / grade group <1 (accuracy = 87 % ) or grade group >1 / grade group ≤1 (accuracy = 91 % ). CONCLUSIONS: In-situ Raman spectroscopy combined with mpMRI radiomics features can lead to highly accurate PCa detection for improved in-vivo targeting of biopsy sample collection and radiotherapy seed placement.

Saliva-based detection of COVID-19 infection in a real-world setting using reagent-free Raman spectroscopy and machine learning.

SIGNIFICANCE: The primary method of COVID-19 detection is reverse transcription polymerase chain reaction (RT-PCR) testing. PCR test sensitivity may decrease as more variants of concern arise and reagents may become less specific to the virus. AIM: We aimed to develop a reagent-free way to detect COVID-19 in a real-world setting with minimal constraints on sample acquisition. The machine learning (ML) models involved could be frequently updated to include spectral information about variants without needing to develop new reagents. APPROACH: We present a workflow for collecting, preparing, and imaging dried saliva supernatant droplets using a non-invasive, label-free technique-Raman spectroscopy-to detect changes in the molecular profile of saliva associated with COVID-19 infection. RESULTS: We used an innovative multiple instance learning-based ML approach and droplet segmentation to analyze droplets. Amongst all confounding factors, we discriminated between COVID-positive and COVID-negative individuals yielding receiver operating coefficient curves with an area under curve (AUC) of 0.8 in both males (79% sensitivity and 75% specificity) and females (84% sensitivity and 64% specificity). Taking the sex of the saliva donor into account increased the AUC by 5%. CONCLUSION: These findings may pave the way for new rapid Raman spectroscopic screening tools for COVID-19 and other infectious diseases.

Time-gated interferometric detection increases Raman scattering to fluorescence signal ratio in biological samples.

Attainable levels of signal-to-background ratio (SBR) in Raman spectroscopy of biological samples is limited by the presence of endogenous fluorophores. It is customary to remove the ubiquitous fluorescence background using postacquisition data processing. However, new approaches are needed to reduce background contributions and maximize the fraction of the sensor dynamical range occupied by Raman photons. Time-resolved detection using pulsed lasers and time-gated measurements can be used to address the signal-to-background problem in biological samples by limiting light detection to nonresonant interaction phenomena with relaxation time scales occurring on sub-nanosecond time scales, thereby excluding contributions from resonant phenomena such as fluorescence. A time-gated Fourier-transform spectrometer was assembled using a commercially available interferometer, a single channel single-photon avalanche diode and time tagging electronics. A time gate of 300 ps increased the signal-to-background-ratio of the 1440 cm-1 Raman band from 36% to 69% in an olive oil sample hereby demonstrating the potential of this approach for autofluorescence suppression.

Multispectral label-free Raman spectroscopy can detect ovarian and endometrial cancer with high accuracy.

Up to 70% of ovarian cancer patients are diagnosed with advanced-stage disease and the degree of cytoreduction is an important survival prognostic factor. The aim of this study was to evaluate if Raman spectroscopy could detect cancer from different organs within the abdominopelvic region, including the ovaries. A Raman spectroscopy probe was used to interrogate specimens from a cohort of nine patients undergoing cytoreductive surgery, including four ovarian cancer patients and three patients with endometrial cancer. A feature-selection algorithm was developed to determine which spectral bands contributed to cancer detection and a machine-learning model was trained. The model could detect cancer using only eight spectral bands. The receiver-operating-characteristic curve had an area-under-the-curve of 0.96, corresponding to an accuracy, a sensitivity and a specificity of 90%, 93% and 88%, respectively. These results provide evidence multispectral Raman spectroscopy could be developed to detect ovarian cancer intraoperatively.

Multispectral intraoperative imaging for the detection of the hemodynamic response to interictal epileptiform discharges.

Interictal epileptiform discharges (IEDs) are brief neuronal discharges occurring between seizures in patients with epilepsy. The characterization of the hemodynamic response function (HRF) specific to IEDs could increase the accuracy of other functional imaging techniques to localize epileptiform activity, including functional near-infrared spectroscopy and functional magnetic resonance imaging. This study evaluated the possibility of using an intraoperative multispectral imaging system combined with electrocorticography (ECoG) to measure the average HRF associated with IEDs in eight patients. Inter-patient variability of the HRF is illustrated in terms of oxygenated hemoglobin peak latency, oxygenated hemoglobin increase/decrease following IEDs, and signal-to-noise ratio. A sub-region was identified using an unsupervised clustering algorithm in three patients that corresponded to the most active area identified by ECoG.

Dimensional reduction based on peak fitting of Raman micro spectroscopy data improves detection of prostate cancer in tissue specimens.

SIGNIFICANCE: Prostate cancer is the most common cancer among men. An accurate diagnosis of its severity at detection plays a major role in improving their survival. Recently, machine learning models using biomarkers identified from Raman micro-spectroscopy discriminated intraductal carcinoma of the prostate (IDC-P) from cancer tissue with a ≥85 % detection accuracy and differentiated high-grade prostatic intraepithelial neoplasia (HGPIN) from IDC-P with a ≥97.8 % accuracy. AIM: To improve the classification performance of machine learning models identifying different types of prostate cancer tissue using a new dimensional reduction technique. APPROACH: A radial basis function (RBF) kernel support vector machine (SVM) model was trained on Raman spectra of prostate tissue from a 272-patient cohort (Centre hospitalier de l'Université de Montréal, CHUM) and tested on two independent cohorts of 76 patients [University Health Network (UHN)] and 135 patients (Centre hospitalier universitaire de Québec-Université Laval, CHUQc-UL). Two types of engineered features were used. Individual intensity features, i.e., Raman signal intensity measured at particular wavelengths and novel Raman spectra fitted peak features consisting of peak heights and widths. RESULTS: Combining engineered features improved classification performance for the three aforementioned classification tasks. The improvements for IDC-P/cancer classification for the UHN and CHUQc-UL testing sets in accuracy, sensitivity, specificity, and area under the curve (AUC) are (numbers in parenthesis are associated with the CHUQc-UL testing set): +4 % (+8 % ), +7 % (+9 % ), +2 % (6%), +9 (+9) with respect to the current best models. Discrimination between HGPIN and IDC-P was also improved in both testing cohorts: +2.2 % (+1.7 % ), +4.5 % (+3.6 % ), +0 % (+0 % ), +2.3 (+0). While no global improvements were obtained for the normal versus cancer classification task [+0 % (-2 % ), +0 % (-3 % ), +2 % (-2 % ), +4 (+3)], the AUC was improved in both testing sets. CONCLUSIONS: Combining individual intensity features and novel Raman fitted peak features, improved the classification performance on two independent and multicenter testing sets in comparison to using only individual intensity features.

Special Series Guest Editorial: Artificial Intelligence and Machine Learning in Biomedical Optics.

Guest editors Behrouz Shabestri, Mark Anastasio, Baowei Fei, and Frédéric Leblond provide an overview of the JBO Special Series on Artificial Intelligence Machine Learning in Biomedical Optics.