Skip to main content

Endoscopic Comprehensive Optical Multimodal Molecular Intelligent Imaging

Periodic Reporting for period 3 - ENCOMOLE-2i (Endoscopic Comprehensive Optical Multimodal Molecular Intelligent Imaging)

Reporting period: 2019-01-01 to 2020-06-30

Seeing or more general sensing with light can be one of the most accurate and precise ways to identify and analyze an object or the environment. In medicine, optical imaging in particular provides highly specific sensing tools to detect the most subtle changes in an organism’s structure, microstructure and metabolism. Hence, optics often is a tool of choice for diagnosis.
The most simple form of optical imaging in medicine is simply looking at the patient for visual diagnosis. However, this also reveals the main problem of optics in medicine, which is the poor penetration of light into the human body, which is a couple of millimetres only. To solve this problem, we saw already 150 years ago the advent of the first endoscopes, mechanical devices introduced into the human body with optics assemblies allowing to look into organs like the stomach.
By now endoscopes are standard tools in medicine and find widespread applications throughout all disciplines. Most of these endoscopes are standard reflection imaging type endoscopes, meaning they provide a simple visual impression of the tissue like if observed with the naked eye. However, modern optics, especially if using lasers, can provide many more imaging modalities and types of contrast, theoretically enabling the design of endoscopes capable of more than just taking a simple color image of tissue. There is multiphoton imaging, hyperspectral imaging, Raman imaging, mid infrared vibrational imaging and many more. Today, hardly any of these advanced imaging devices can be found in endoscopes, even so they might be much more powerful to detect diseases earlier, more reliably, more sensitively, and more specifically. This problem is addressed in the ERC project ENCOMOLE-2i, where advanced endoscope based imaging systems are developed which are capable of rapidly scanning large areas of tissue with molecular contrast, high sensitivity and full 3D capability. To take full advantage of the wealth of information and the wealth of possibilities, these systems will provide an artificial intelligence based control unit to adaptively adjust the system for optimum imaging strategies.
We hope that these new advanced endoscopic intelligent imaging systems will enable to detect the onset of disease earlier and more precisely. This will help to more efficiently treat diseases, because in many cases the earlier a pathologic condition is diagnosed the better are the chances for effective treatment. Maybe patients at risk can even be identified and diagnosed before the onset of disease symptoms. This can relate to early cancer diagnosis but also to other diseases like inflammatory disorders. Such a system can contribute to a more personalized more individualised therapy for future precision medicine.
Considering that optics in form of microscopes or endoscopes is used for all groups of major diseases, i.e. cardiovascular diseases, cancer and infectious diseases, the impact of advanced optical systems for in vivo imaging inside patients cannot be underestimated. Not only with respect to our ageing society the earlier diagnosis of disease and the improved management and treatment can contribute to a substantially decreased cost of health care but more importantly it may contribute to improved health and wellbeing of many millions of citizens.
The first part of the ERC project ENCOMOLE-2i focused on designing novel endoscopes, appropriate for the advanced optical imaging modalities to be used in the future system and it focused on on finalising, optimising and adopting three advanced optical imaging modalities to be integrated into endoscopes.

The frist research track within ENCOMOLE-2i dealt with developing appropriate means to deliver the light into the human body. With respect to new developments of endoscopes, we were able to evaluate three different, novel endoscope designs and interfaced them with our MHz-OCT. First, these were rigid endoscopes where we could achieve real-time 4D live optical coherence tomography with imaging speeds of more than 25 full volumes per second. The second design, in collaboration with the group of Prof. G. Hüttmann (BMO, Lübeck), was a piezo electric transducer based forward looking spiral scan endoscope which was also capable of acquiring OCT data very rapidly, here at a rate of 6 volumes per second. The third design, developed in the group of T. Wang and G. V. Soest (Erasmus M.C. Rotterdam), was based on an electric micro motor capable of acquiring more than 5600 frames per second. By interfacing these endoscopes with our OCT-engines, we were able to achieve world record imaging speeds with all three different designs. This fulfils the requirement in ENCOMOLE-2i for an endoscopic OCT system, fast enough to provide live navigation through tissue in order to guide the other imaging modalities integrated into the same endoscope.

The second research track focused on the 3 imaging modalities to be used with our advanced endoscopes. These were: (1) MHz-OCT, i.e. optical coherence tomography with line rates of more than 1 million depth scans per second corresponding to pixel rates of more than 1 billion elements per second with the requirement that the system is capable of acquiring, processing, analysing and displaying the resulting data streams on the order of 8 gigabytes per second in real time and live without noticeable delay. This is the prerequisite to use OCT as online imaging modality for navigation and for live feedback. (2) The second imaging modality is stimulated Raman sensing using the OCT laser to detect and identify the biomolecules present in a sample. This molecular imaging technique is slower but it is capable of providing label-free endogeneous contrast of the molecular composition of a sample. Stimulated Raman sensing can be extremely sensitive and powerful for the detection of early pathological changes of tissue caused by the onset of disease. However, stimulated Raman sensing, especially if it is performed in the highly desired fingerprint region, can be slow. For this reason we suggested to apply special, smart, sparse spectral sampling in order to find the optimum compromise between speed, resolution and signal strength. We successfully developed and published this technique. (3) The third imaging modality, which will be integrated into our multimodal intelligent endoscope is 2-photon microscopy using sub-nanosecond laser pulses for excitation. The comparably long pulse duration suppresses self-phase modulation and pulse breakup in optical fibres as a prerequisite to make 2-photo fiber endoscopes more feasible. We developed, implemented and demonstrated such a sub ns-2-photon engine and could also show that the longer duration of the laser pulses is also enabling a new way to detect fluorescence lifetime signals. Enabled by high-speed analogue to digital converters, which just recently became available, the direct real time sampling of the fluorescence lifetime signal is possible. This enabled us to demonstrate two-photon fluorescence lifetime imaging at a record speed of more than 1 million lifetime traces per second. It is noteworthy to point out that the fluorescence lifetime signal is automatically always measured in parallel with our two-photon fluorescence signal. This means our system always acquires additional information from the sample which might be useful to increase the sensitivity and precision of our imaging system even more.

So in summary we were able to develop new appropriate endoscope devices, improve our imaging engines with the three different imaging modalities and even discover a new way to add additional functionally sensitive contrast.
We demonstrated the world’s fastest endoscopic OCT with real-time live processing. With 25 volumes per second for rigid endoscopes and 6 volumes per second for forward looking spiral scan fiber endoscopes and a line rate of 3 MHz and a voxel rate of 2 billion points per second this is by far the fastest endoscopic live OCT ever demonstrated. This performance enables us to have an endoscope which can be interactively used purely guided by the OCT signal itself. It will serve as a foundation stone to have a multimodal advanced optical endoscope which can be used interactively. To the best of our knowlege, the 1 MHz pixel rate, corresponding to a 1µs pixel dwell time, that we demonstrated with our 2-photon fluorescence lifetime system can also be up to 10x faster than the fastest TCSP-counting systems reported to date.