Wavelength weightings in machine learning for ovine joint tissue differentiation using diffuse reflectance spectroscopy (DRS).

Objective: To investigate the DRS of ovine joint tissue to determine the optimal optical wavelengths for tissue differentiation and relate these wavelengths to the biomolecular composition of tissues. In this study, we combine machine learning with DRS for tissue classification and then look further at the weighting matrix of the classifier to further understand the key differentiating features. Methods: Supervised machine learning was used to analyse DRS data. After normalising the data, dimension reduction was achieved through multiclass Fisher's linear discriminant analysis (Multiclass FLDA) and classified with linear discriminant analysis (LDA). The classifier was first run with all the tissue types and the wavelength range 190 nm - 1081 nm. We analysed the weighting matrix of the classifier and then ran the classifier again, the first time using the ten highest weighted wavelengths and the second using only the single highest. Our method was applied to a dataset containing ovine joint tissue including cartilage, cortical and subchondral bone, fat, ligament, meniscus, and muscle. Results: It achieved a classification accuracy of 100% using the wavelength 190 nm - 1081 nm (2048 attributes) with an accuracy of 90% being present for 10 attributes with the exception of those with comparable compositions such as ligament and meniscus. An accuracy greater than 70% was achieved using a single wavelength, with the same exceptions. Conclusion: Multiclass FLDA combined with LDA is a viable technique for tissue identification from DRS data. The majority of differentiating features existed within the wavelength ranges 370-470 and 800-1010 nm. Focusing on key spectral regions means that a spectrometer with a narrower range can potentially be used, with less computational power needed for subsequent analysis.

Machine learning classification of human joint tissue from diffuse reflectance spectroscopy data.

Objective: To assess if incorporation of DRS sensing into real-time robotic surgery systems has merit. DRS as a technology is relatively simple, cost-effective and provides a non-contact approach to tissue differentiation. Methods: Supervised machine learning analysis of diffuse reflectance spectra was performed to classify human joint tissue that was collected from surgical procedures. Results: We have used supervised machine learning in the classification of a DRS human joint tissue data set and achieved classification accuracy in excess of 99%. Sensitivity for the various classes were; cartilage 99.7%, subchondral 99.2%, meniscus 100% and cancellous 100%. Full wavelength range is required for maximum classification accuracy. The wavelength resolution must be larger than 8nm. A SNR better than 10:1 was required to achieve a classification accuracy greater than 50%. The 800-900nm wavelength range gave the greatest accuracy amongst those investigated Conclusion: DRS is a viable method for differentiating human joint tissue and has the potential to be incorporated into robotic orthopaedic surgery.

Excitability and optical pulse generation in semiconductor lasers driven by resonant tunneling diode photo-detectors.

We demonstrate, experimentally and theoretically, excitable nanosecond optical pulses in optoelectronic integrated circuits operating at telecommunication wavelengths (1550 nm) comprising a nanoscale double barrier quantum well resonant tunneling diode (RTD) photo-detector driving a laser diode (LD). When perturbed either electrically or optically by an input signal above a certain threshold, the optoelectronic circuit generates short electrical and optical excitable pulses mimicking the spiking behavior of biological neurons. Interestingly, the asymmetric nonlinear characteristic of the RTD-LD allows for two different regimes where one obtain either single pulses or a burst of multiple pulses. The high-speed excitable response capabilities are promising for neurally inspired information applications in photonics.

Photo-detectors integrated with resonant tunneling diodes.

We report on photo-detectors consisting of an optical waveguide that incorporates a resonant tunneling diode (RTD). Operating at wavelengths around 1.55 µm in the optical communications C band we achieve maximum sensitivities of around 0.29 A/W which is dependent on the bias voltage. This is due to the nature of RTD nonlinear current-voltage characteristic that has a negative differential resistance (NDR) region. The resonant tunneling diode photo-detector (RTD-PD) can be operated in either non-oscillating or oscillating regimes depending on the bias voltage quiescent point. The oscillating regime is apparent when the RTD-PD is biased in the NDR region giving rise to electrical gain and microwave self-sustained oscillations Taking advantage of the RTD's NDR distinctive characteristics, we demonstrate efficient detection of gigahertz (GHz) modulated optical carriers and optical control of a RTD GHz oscillator. RTD-PD based devices can have applications in generation and optical control of GHz low-phase noise oscillators, clock recovery systems, and fiber optic enabled radio frequency communication systems.

Integrated microspectrometer for fluorescence based analysis in a microfluidic format.

We have demonstrated a monolithic integrated arrayed waveguide grating (AWG) microspectrometer microfluidic platform capable of fluorescence spectroscopic analysis. The microspectrometer in this proof of concept study has a small (1 cm × 1 cm) footprint and 8 output channels centred on different wavelengths. We show that the signals from the output channels detected on a camera chip can be used to recreate the complete fluorescence spectrum of an analyte. By making fluorescence measurements of (i) mixed quantum dot solutions, (ii) an organic fluorophore (Cy5) and (iii) the propidium iodide (PI)-DNA assay, we illustrate the unique advantages of the AWG platform for simultaneous, quantitative multiplex detection and its capability to detect small spectroscopic shifts. Although the current system is designed for fluorescence spectroscopic analysis, in principle, it can be implemented for other types of analysis, such as Raman spectroscopy. Fabricated using established semiconductor industry methods, this miniaturised platform holds great potential to create a handheld, low cost biosensor with versatile detection capability.

High frequency optoelectronic oscillators based on the optical feedback of semiconductor mode-locked laser diodes.

Optical self seeding feedback techniques can be used to improve the noise characteristics of passively mode-locked laser diodes. External cavities such as fiber optic cables can increase the memory of the phase and subsequently improve the timing jitter. In this work, an improved optical feedback architecture is proposed using an optical fiber loop delay as a cavity extension of the mode-locked laser. We investigate the effect of the noise reduction as a function of the loop length and feedback power. The well known composite cavity technique is also implemented for suppressing supermode noise artifacts presented due to harmonic mode locking effects. Using this method, we achieve a record low radio frequency linewidth of 192 Hz for any high frequency (>1 GHz) passively mode-locked laser to date (to the best of the authors' knowledge), making it promising for the development of high frequency optoelectronic oscillators.

Output Power Limitations and Improvements in Passively Mode Locked GaAs/AlGaAs Quantum Well Lasers.

We report a novel approach for increasing the output power in passively mode locked semiconductor lasers. Our approach uses epitaxial structures with an optical trap in the bottom cladding that enlarges the vertical mode size to scale the pulse saturation energy. With this approach we demonstrate a very high peak power of 9.8 W per facet, at a repetition rate of 6.8 GHz and with pulse duration of 0.71 ps. In particular, we compare two GaAs/AlGaAs epilayer designs, a double quantum well design operating at 830 nm and a single quantum well design operating at 795 nm, with vertical mode sizes of 0.5 and 0.75 µm, respectively. We show that a larger mode size not only shifts the mode locking regime of operation towards higher powers, but also produces other improvements in respect of two main failure mechanisms that limit the output power: the catastrophic optical mirror damage and the catastrophic optical saturable absorber damage. For the 830 nm material structure, we also investigate the effect of non-absorbing mirrors on output power and mode locked operation of colliding pulse mode locked lasers.

A Self-Synchronized Optoelectronic Oscillator based on an RTD Photo-Detector and a Laser Diode.

We propose and demonstrate a simple and stable low-phase noise optoelectronic oscillator (OEO) that uses a laser diode, an optical fiber delay line and a resonant tunneling diode (RTD) free-running oscillator that is monolithic integrated with a waveguide photo-detector. The RTD-OEO exhibits single-side band phase noise power below -100 dBc/Hz with more than 30 dB noise suppression at 10 kHz from the center free-running frequency for fiber loop lengths around 1.2 km. The oscillator power consumption is below 0.55 W, and can be controlled either by the injected optical power or the fiber delay line. The RTD-OEO stability is achieved without using other high-speed optical/optoelectronic components and amplification.