Detector Breakthroughs

To perform time-resolved fluorescence imaging in the lung at video rates it is vital for our optical sensors to incorporate fast on-chip signal processing.

Two of the most important areas where we have achieved world-record data throughput rates are in time-to-digital conversion and on-chip histogramming.

Time-to-Digital Conversion (TDC)

Time to digital conversion is the process of measuring time intervals by means of electronic circuitry and representing time as a digital number or code which can be processed or stored. The conversion is performed by a category of circuits known as time to digital converters (TDCs) which are intricate systems capable of measuring time intervals typically with tens of picoseconds of resolution.

Such circuits require two main trigger signals for them to operate, a start signal indicating the start of the time interval to be measured, and a stop signal which halts the measurement resulting in a representative digital output.

An equivalent analogy to these circuits is the stopwatch which is used to measure fine time intervals between the start of a race and the time an athlete crosses the finish line. To put this in context consider the example above, where a laser is placed a certain distance away from a photo-detector with both of them connected to a TDC circuit. The start signal is a laser trigger indicating that a laser pulse has been generated and the athlete is the photon that is racing towards the finish line or the photo-detector. Once the photo-detector or SPAD detects the photon, it generates the stop signal causing the TDC or the stopwatch to record the time of arrival with respect to the laser start.

On-Chip Histogramming

A TDC is capable of generating an enormous number of photon time-stamps representing a significant data readout, storage and analysis challenge. Often the real quantity of interest in experiments is the histogram showing the distribution of photon arrival times. Typically a time-resolved histogram is required for each wavelength under investigation. From these histograms the characteristic time-constants associated with particular fluorophores may be evaluated, and color-maps corresponding to these time constants can be used to create a powerful rendering of lung images where different cells and molecules can be readily distinguished.

Histograms may be computed in a post-processing stage from raw time-stamps stored on the host-computer. The disadvantage here is that the histograms may take a long time to compute for large datasets and the ability to navigate and analyze in the lung in real-time is lost to the clinician. Opportunities to gain significant information about lung pathogens and conditions may therefore be lost. A better approach is to compute histograms in the FPGA controller but this is often challenging in terms of  programmable digital logic resources. The best approach is to generate histograms directly at source – on the sensor-chip itself using hard-wired ultrafast digital logic. This ‘system-on-a-chip’ approach is used in the Proteus RAII sensor.