Stereotactic needle biopsy is a time-consuming and invasive procedure that often cannot accurately distinguish recurrent tumors from treatment effect in gliomas. We report an intraoperative multispectral fluorescence lifetime imaging (FLIm) system coupled with a custom-made fiber optic probe integrated with the stealth biopsy needle as an optical biopsy tool. FLIm parameters collected from 3 suspected recurrent glioma patients changed over the biopsy trajectory as the needle passed different brain areas. An SVM classifier validated using a leave-one-patient-out validation scheme could identify the lesions from the normal surrounding tissue with sensitivity=0.99, specificity=0.91, and accuracy=0.95.

Accurate detection of brain tumor boundaries is crucial for successful tumor removal and better patient outcomes. A novel method using label-free Fluorescence Lifetime Imaging (iFLIm) is presented in this study. The approach involved developing an optimized classification model based on tumor enhancement status, utilizing multispectral FLIm. The method was evaluated on 52 patients with adult-type diffuse glioma, demonstrating promising results with 87% sensitivity, 92% specificity, and an AUC of 0.90. This FLIm-based model has the potential to offer a non-invasive and real-time technique to assist neurosurgeons in accurately identifying tumor infiltrates, potentially improving tumor resection and patient outcomes.

Deep brain stimulation (DBS) surgery is performed on patients suffering Parkinson’s disease for whom medication is no longer effective in relieving their motor symptoms. In this surgery, a stimulating electrode is implanted in a specific structure deep within the brain, delivering electrical impulses and thus reducing the motor symptoms. The success of the surgery is highly dependent on placing the electrode accurately in the targeted structure, typically the subthalamic nucleus (STN). We developed a DBS electrode that includes optical fibers to perform coherent anti-Stokes Raman scattering (CARS) spectroscopy and diffuse reflectance spectroscopy (DRS) during the electrode insertion in the brain. We were able to identify white and grey matter using principal component analysis (PCA), showing that spectroscopic measurements could be suitable for neuronavigation.

This research aims to deal with intraoperative multispectral images taken from brain tumour surgeries to investigate the diagnostic and guidance potential of MSI. These images were registered by feature-based (SIFT, PFN), intensity-based (LK) and machine learning (RANSAC-Flow) methods and classified via a CNN and Transformer model using anatomical labels. Based on the results from some initial training, MSI could achieve 95% overall accuracy. After labelling and registration are completed, a brain surgery dataset can be built to support intraoperative decision making.

In this study, we investigated the use of single-point Raman spectroscopy (RS) to differentiate between normal brain tissue and focal cortical dysplasia (FCD) during epilepsy surgery. Initial RS measurements showed high background noise, but using a different Raman probe with fewer optic fibers, we were able to detect characteristic brain spectra bands in three patients. These bands confirmed the optical signature of normal brain tissue. While statistical significance was not achieved due to the limited number of patients, our study improved the technical methodology for processing brain signals and holds promise for intraoperative detection of FCD in epilepsy surgeries. Further research is needed to validate these findings and enhance surgical outcomes.

We compared the diagnostic accuracy of label-free Photon Absorption Remote Sensing (PARS) virtual hematoxylin and eosin (H&E) staining to conventional histology in breast cancer core needle biopsy (CNB) samples. Matching PARS virtual H&E and real H&E images were acquired for seven breast CNB samples representing a spectrum of histologies. Seven board certified pathologists, blinded to the image origin, provided diagnostic evaluations of each sample. A kappa concordance analysis was performed and revealed 'substantial' agreement (kappa > 0.6) between the PARS and H&E pairs in primary diagnosis. Pathologists were unable to distinguish the true origin of the PARS virtual H&E images.

Traditional diagnostic methods for burn wounds have remained inaccurate.We have recently demonstrated that terahertz (THz) spectroscopic imaging can assess wound severity and predict healing outcome with high accuracy using our Portable HAndheld Spectral Reflection (PHASR) Scanners, which provide fast full-spectroscopic imaging. We will describe recent work exploring different physics-based machine learning methods to classify wounds using THz spectral data. THz images captured 1-hour post-burn achieve an accuracy of 94.7% in predicting the wound healing outcome by 28 days. A reduced-dimensionality double-Debye model describes the refractive index of the tissue over the entire spectra using only five empirical parameters. A neural network based on this model still achieved 88% healing outcome prediction accuracy. Finally, we will discuss plans to translate this technology to clinical trials.

A novel human-machine interface (HMI) combining mobile optical coherence tomography (OCT) and deep learning algorithms enables automatic identification of lung lesions during surgery. With over 80% sensitivity and specificity, this technique facilitates rapid histologically graded diagnosis, providing fast information to clinicians. It offers a cost-effective approach for early detection and treatment guidance, benefiting patients and advocating their rights in the battle against lung cancer.

During the epidural injection procedure, identifying the exact tissue is essential. Physicians typically use the loss of resistance (LOR) to confirm the penetration of the needle into the epidural space. However, this navigation task can be challenging. The optical sensor system has the advantages of high sensitivity, resolution, and miniaturization, so it is suitable for precision biomedical applications. In this study, we presented a new FPI sensor system that simultaneously measures force and distance and an algorithm for LOR. The needle insertion experiments conducted by the user showed that the LOR success rate increased using the proposed system.

A technologic solution has been developed to stream 3D stereoscopic videos from the operating microscope in real-time with minimal latency to any place in the world for remote advice from colleagues or supervision of trainees. From the ocular views, side-by-side 3D videos were constructed and streamed through a secure link to a mobile phone placed in VR glasses for 3D viewing with minimal latency (<0.5s). The setup was successfully tested during otologic surgical procedures. This remote 3D viewing solution proves to be safe, practical and effective for expert assistance and education during for microsurgeries while fulfilling applicable regulations.

Detecting oesophageal cancer early leads to significantly better patient outcomes. Patients with conditions associated with higher cancer risk undergo regular surveillance endoscopies yet due to low contrast in white light imaging, standard of care endoscopies suffer from low detection rates. Hyperspectral imaging can improve contrast between healthy and diseased tissue. Here, we developed a system enabling flexible illumination-based hyperspectral imaging, designed to be used with a standard clinical endoscope. Our custom light source consists of white light from a supercontinuum laser and one of two tunable illumination systems: an acousto-optic tunable filter, enabling narrowband spectral sweeps, or a high-speed digital micromirror device, resulting in a highly versatile system for clinical testing.

A pilot study on 20 normal-risk volunteers was performed to evaluate the safety and feasibility of the falloposcope, an experimental microendoscope for imaging the fallopian tubes (FTs) that contains optical coherence tomography (OCT) and multispectral fluorescence imaging (MFI). In vivo and ex vivo imaging of the fallopian tubes (FTs) was performed. No reportable events occurred and only non-significant FT disruption was noted on subsequent histology. Usable MFI data were obtained for approximately half of the participants, and OCT images were obtained for 12 participants. Future studies will evaluate the diagnostic potential of the falloposcope.

Surgeries performed on human eye require a high level of dexterity. Accuracy of widely used ophthalmic stereomicroscopes can be affected by limited depth perception and precision reduced by physiological tremor. Robotic arm-assisted manipulation of surgical tools improves precision, while intraoperative Optical Coherence Tomography (OCT) provides depth information and improves accuracy. Our hardware and software solutions enable tracking the tool tip to ensure the displayed OCT cross-scans, rather than 3-D volumes, are always locked to the tools’ tip position parsed from the robotic arm subsystem. Tool tracking errors for lateral movement and rotation are discussed for various velocities of the tool.

Here we proposed a designed, built, and evaluated 20G vertically inserted razor edge cannula (VIREC) robotic device guided by optical coherence tomography (OCT) for pneumatic dissection. The fiber sensor was glued inside the needle at a fixed offset of ~500 um. During the experiment, the robotic needle driver precisely moves the VIREC based on the surgeon input which is carefully monitored by the M-mode OCT system. Once the needle is inserted into the desired depth, the air is injected by the surgeon to separate stroma from Descemet’s membrane (DM). During in vivo study (N=8), the “big bubble” was effectively generated in six of eight eyes tested and DM was perforated in two eyes. This demonstrated the reliability and effectiveness of VIREC for “big bubble” DALK.

Prostate cancer represents 7% of all new cancer cases worldwide. There remain important challenges in the diagnosis and treatment of prostate cancers, particularly to increase biopsy accuracy and to enable nerve-sparing surgery. Here, we explore the capability of wide-field optical coherence tomography (OCT) and optical coherence elastography (OCE), spanning up to 45x45 mm in the lateral field of view and to depths of 1 mm, to accurately identify prostate microarchitecture. In this study, we present novel methodology to systematically validate OCT and OCE images through co-registration with standard histology on 15 human prostate specimens and demonstrate the potential for intraoperative use.

Single fiber reflectance spectroscopy can measure tissue optical properties and analyze physiological state. However, the standard implementation of SFR comprises several sources of instabilities that are typically not accounted for. These are fiber modal content, instabilities in the illumination spectrum, and bend losses. These instabilities lead to reduced reproducibility and reliability. We propose a new SFR system design with dynamic correction of source fluctuations and bend losses. Furthermore, we replace the traditional bifurcated fiber with a wideband multimode circulator for improved light transmission and reduced modal sensitivity. We demonstrate the enhanced stability in a one-to-one comparison with a standard system.

This study integrates Diffuse Reflectance Spectroscopy (DRS) into spine surgery by incorporating optical fibers into awls, commonly used for pedicle cannulation. The sharp awl tip poses a challenge for incorporating parallel fibers at the distal end. Through Monte Carlo simulations and optical phantom experiments, we assess the influence of the extended tip on the optical signal and explore alternative slanted tip designs. Results show that the tip design affects the probed volume. While the extended tip impedes the detection of cortical breaches, the slanted layout maintains a consistent look-ahead distance beyond the awl tip. These important insights highlight the potential of slanted fiber-optic awls for accurate spine surgery.

We present a fully integrated, clinical-compatible SRS imaging device giving access to the complete Raman spectrum during tumor surgeries. For detailed examination of larger tissue samples, we have designed a high-speed, low-resolution imaging mode to quickly identify cancerous hot-spots, followed by a high-resolution imaging mode. Leveraging the advantages of a compact and robust fiber laser, we have integrated the entire microscopy system into a clinical cart, facilitating deployment in diverse clinical environments. The laser provides rapid tunability within milliseconds across a broad spectral range of 700 to 3300 1/cm, covering biomedically relevant resonances in the fingerprint region.

For the development and industrialization of endoscope systems, the miniaturization of high-performance optics, electronics and light sources at a reasonable pricing and a fast time-to-market are the biggest challenges. It is success-relevant to have the right industry partner on hand to support state-of-the-art 4k stereovision in smallest diameters including multimodal approaches (multispectral, IR fluorescence, stereoscopy) and full ISO13485 certified product lifecycle management - from the initial product idea to high volume clean room manufacturing. Cross functional expertise is the key to develop reliable endoscopes and light modules, chip-in-tip or relay optics, electronics and firmware with a high degree of customization. A stable and reliable supply chain for a broad range of high-tech components and in deep IP competence back up the all-inclusive development.

Complete blood cell count (CBC) is one of the most valuable clinical tests to measure alternations in leukocyte numbers and to provide signs of infection and severe inflammation (e.g., sepsis). However, the CBC procedure is invasive, often requiring multiple blood samples for ex vivo laboratory analysis. Moreover, it does not provide dynamic cell information such as enhanced adhesiveness of leukocytes on the blood vessel walls, a cardinal feature of inflammation. We have developed a miniaturized oblique back-illumination microscope capable of noninvasive, high-resolution and label-free imaging of human microvasculature. The instrument was used to image and quantify circulating leukocytes, and the results are compared with CBC. We also investigated the leukocyte-endothelial interactions in inflamed conditions, revealing a drastic change in cell rolling speed. To provide clinicians with reliable and actionable results, we are developing a deep learning algorithm for automated quantification and movement analysis.