Conference Agenda

Non-invasive, real-time venous oxygen saturation (SvO2) measurements provide the potential to improve health outcomes in transfusions, ventilator care, and in the ICU. Current measurements involve catheters which are invasive and expensive with potential high-risk, providing discrete measurements from a jugular vein. This research designed, developed, and tested a proof-of-concept optical sensor similar to a pulse oximeter for non-invasive, continuous SvO2 monitoring at the external jugular vein. Subject testing met the aim of correctly identifying the EJV waveform and provided an SvO2 estimation within the usual range of 60-90%. SvO2 estimates from the EJV pulse of 71.14%, 72.21%, and 70.41% and breathing pulse of 74.79%, 75.32%, and 74.12%, all fall within the range, with trials capturing clear EJV waveforms from the single-point sensor. Further research is necessary to calibrate and validate the device against gold-standard blood gas analysers. However, the initial results confirm the sensor’s reliability and potential in detecting and estimating SvO2 non-invasively, which has the potential to benefit patients across a wide range of clinical settings.

Introduction

Breast conserving resection with free margins for the gold standard treatment of early breast cancer requires a reliable discrimination between normal and malignant tissue at the resection margins. In this study, normal and abnormal tis-sue samples from breast cancer patients were characterized ex vivo by optical emission spectroscopy (OES). The aim of the study was to determine spectroscopic features which allow future real-time tissue differentiation for margin assess-ment during breast cancer surgery.

Methods

Optical emission spectra were recorded during electrosurgical treatment of ex vivo tissue samples from breast cancer surgery of patients with ≥15mm invasive breast tumors of all histologic subtypes, but without neoadjuvant therapy. After histologic assignment of the tissue samples to either normal or abnormal tissue and its correlation with the corre-sponding spectra, the classification performance for the differentiation of normal vs. abnormal breast tissue by ma-chine learning algorithms was evaluated.

Results

A data set of 972 spectra, comprising 480 from normal and 492 from abnormal tissues, was used for analysis. Spectro-scopic features were detected and identified by a custom-designed analysis algorithm, and selected according to their occurrence rate in all spectra.

Using all selected spectroscopic features a support vector classifier for tissue discrimination was trained. The classifica-tion of normal and abnormal tissue exhibited high accuracies for each individual patient, resulting in an average accu-racy, sensitivity and specificity of 96.9%, 94.8% and 99.0%, respectively.

Conclusion

Our study demonstrates the potential of RF-OES technology as a promising tool for intraoperative discrimination be-tween benign and malignant breast tissue with high sensitivity, specificity, PPV and NPV due to specific spectroscopic features. Further research is required for the clinical transfer of OES as a unique fully integrated and real-time margin detection method for breast cancer surgery.

Introduction

Optical spectroscopy is an analytical technique in the biomedical field which often relies on expensive and inflexible solutions with a high level of equipment complexity. Especially due to the increased need for real-time, on-site data ac-quisition in biomedical research and diagnostics, miniaturization of such systems (lab-on-chip) would enable cost-effective real-time sensor technology. This motivates the development and construction of a novel micro-spectrometer system for the near-infrared (NIR) range. NIR spectroscopy is analytical technique that is particularly suitable for or-ganic components that requires less sample preparation and thus becoming increasingly important in biomedicial anal-ysis and diagnostics. A limiting factor for the broad application of NIR spectroscopy are the available detectors for wavelengths above 1000 nm on the market: indium gallium arsenide (InGaAs) detectors are mainly used due to their performance, but at a high fabrication costs. Instead we use germanium-on-silicon (Ge-on-Si) detectors which are fab-ricated by standard semiconductor processes, can be monolithically integrated and scaled to larger wafer sizes, which makes it significantly more cost-effective than InGaAs- detectors.

Methods

The core of the micro system consists of a spectral sensitive photonic chip that was developed by Institut für Mikrolek-tronik Stuttgart (IMS CHIPS) and Institute of Semiconductor Engineering of the University Stuttgart (IHT). It is equipped with a monolithic germanium-on-silicon (Ge-on-Si) sensor line and a backside diffraction grating which ena-bles spectral resolution in the range of 1000 nm to 1600 nm on a single chip. The Chip Spectrometer is integrated into a highly miniaturized spectroscopy setup for quantitative analysis of a biomedical sample. The focus is on the realization of a simple, cost-effective systemintegration with all necessary system components that stands out from the high-priced, inflexible currently solutions on the market, which allows direct automated detection of an analyte.

Results

For the development of the micro system all spectroscopy components like the photonics chip, a broadband NIR light source, optical paths, sample adaption and electronics were realized under the aspects of cost efficiency, miniaturiza-tion and the aim of a high information content. All components where integrated in the spectroscopy system that is able to spectrally analyse a sample in parallel over the entire spectral sensitivity range (1000 nm to 1600 nm). As proof-of-concept a first table top system on the scale of 100 mm x 80 mm x 100 mm was realized with the potential to be-come significantly smaller. A NIR spectroscopy based quantitative analysis of first liquid samples, different mixtures of water isopropanol in a reflectance and a transmission setup was demonstrated.

Conclusion

A first setup of the micro-spectrometer system was manufactured and successfully tested. This proof-of-concept shows that affortable micro NIR spectrometers for a broad field of applications like biomedicine with a high need of miniatur-ization can be achieved. The next steps in the development would be to further miniaturize the setup in the direction of a chip-scale spectrometer and to address a specific medical application.

Introduction

In the research of Intraoperative Gamma Cameras (IGCs), coded aperture collimators remain disfavoured despite their theoretical benefits. In Coded Aperture Imaging (CAI), collimators are used that consist of multiple pinholes to combine the high sensitivity of parallel-hole collimators with the excellent resolution and magnification prop-erties of pinhole collimators. However, they require image reconstruction. The goal of this presentation is to demonstrate and discuss the potential and the challenges in CAI.

Methods

After previously developing “ConvSim”, a fast simulation framework for CAI with emphasis on nearfield effects, a Convolutional Neural Network (CNN) was trained and compared to other planar reconstruction methods. Fur-ther, 3D imaging was investigated by generating 3D reconstructions from a sequence of planar reconstructions where the in-focus plane was incrementally shifted from 20mm to 100mm. The 3D reconstructions were used to determine both the axial resolution as well as the localisation accuracy for point-like sources that we captured with our experimental IGC.

Results

In planar reconstruction, the CNN approach delivered images of higher quality than analytical methods. Investi-gating 3D imaging showed that it is possible to localize point-like sources with an accuracy that is similar to ste-reo cameras. However, we revealed that currently no reconstruction method exists that is fast, accurate and pre-cise in all three dimensions. Additionally, reconstruction of point-like sources works well, but as the size of the source increases, the reconstruction quality drops dramatically. Related to this is the question of the dependency between the quality and the number of photons collected. Finally, we want to encourage other researchers to share their datasets to provide more groups access to this research topic.

Conclusion

Before IGC can benefit from the good trade-off between sensitivity and spatial resolution of coded aperture col-limators, the challenges described above need to be addressed and possible solutions or limitations investigated.

Catheter-associated urinary tract infections are among the most common hospital-acquired infections and all previous measures to reduce their incidence have been less successful. Therefore, the present study investigates an approach based on the antimicrobial effect of visible light. The radiation of a violet or blue LED is guided via optical fiber with a diffuser tip into a catheter surrounded by an E. coli suspension (phosphate buffered saline or urine) in a simple urinary tract model. The irradiation led to bacterial reductions of up to several orders of magnitude that however depend on various parameters. For example, violet light had a stronger antimicrobial effect than blue light, even in the experiments with urine, which absorbs violet light more strongly than blue light. An increase in irradiation intensity led to a greater reduction in bacteria. Therefore, the application of violet light to prevent catheter-associated urinary tract infections seems particularly promising, even for increased intensity but must be tested on even more realistic models in the future, which should not only include planktonic cells but also biofilms.