Nuclear Instruments and Methods in Physics Research A 477 (2002) 431–434

Automatic scanning for nuclear emulsion

Nicola D’Ambrosio*,1

University of M .unster, Institut f .ur Kernphysik, Wilhelm Klemm Strasse 9, 48149 M .unster, Germany

Abstract

Automatic scanning systems have been recently developed for application in neutrino experiments exploiting nuclear

emulsion detectors of particle tracks. These systems speed up substantially the analysis of events in emulsion, allowingthe realisation of experiments with unprecedented statistics. The pioneering work on automatic scanning has been doneby the University of Nagoya (Japan). The so called new track selector has a very good reproducibility in position(B1 mm) and angle (B3mrad), with the possibility to reconstruct, in about 3 s, all the tracks in a view of 150 � 150 mm2

and 1 mm of thickness. A new system (ultratrack selector), with speed higher by one order of magnitude, has started tobe in operation. R&D programs are going on in Nagoya and in other laboratories for new systems. The scanning speedin nuclear emulsion be further increased by an order of magnitude. The recent progress in the technology of digital

signal processing and of image acquisition systems (CCDs and fast frame grabbers) allows the realisation of systemswith high performance.

New interesting applications of the technique in other fields (e.g. in biophysics) have recently been envisaged. r 2002

Elsevier Science B.V. All rights reserved.

PACS: 14.60.Pq

Keywords: Nuclear emulsion; Automatic scanning; Neutrino oscillation

1. Introduction

Nuclear emulsion detectors provide three-di-mensional spatial information on particle tracks,with excellent resolution (of the order of 1 mm), aswell as high hit density (300 hits/mm) along tracks.They are, therefore, ideal for the unambiguousdetection of short-lived particles. This is crucial toattain the required sensitivity in the CHORUS [1]and OPERA [2] experiments. The feasibility of

these experiments, is linked to the impressiveprogress recently done in the field of computercontrolled microscope, with Charged CoupledDevices (CCD) readout, and in automatic patternrecognition.

2. Nuclear emulsion

Nuclear emulsions are made of micro-crystals ofsilver halides (AgBr) as sensitive devices dispersedin a thin gelatine layer. The size of micro-crystalsranges from 0.2 to 0:3 mm and the concentration ofthe crystals in the emulsion ranges from B25 to50% in volume. The dimension and the type ofcrystals can be controlled by current advanced

*Corresponding author. Present address: National Institute

for Nuclear Physics – Naples (INFN Napoli), via Cintia, Monte

S. Angelo Ed. G, 80131 Napoli, Italy. Tel.: +39-081-676311-14.

E-mail address: nicola.d’ambrosio@cern.ch (N. D’Ambrosio).1From the CHORUS collaboration.

0168-9002/02/$ - see front matter r 2002 Elsevier Science B.V. All rights reserved.

PII: S 0 1 6 8 - 9 0 0 2 ( 0 1 ) 0 1 7 9 0 - 9

industrial technologies developed for nuclearemulsion.

Emulsion Sheets (ES) have emulsion layers onboth sides of a transparent plastic base whosethickness ranges from 70 to 1000 mm. The thick-ness of the emulsion layer can vary from 50 to500 mm, depending on the specific use.

The incident particle beam has usually adirection perpendicular to the ES surface. Whenan ionizing particle traverses the emulsion layer ofan ES, electron–hole pairs are created in thecrystals. Clusters of several silver atoms areformed in several micro-crystals by successiveionic processes. Chemical processing is requiredto make the ES readable by scanning systems. Bychemical action the small silver clusters (calledlatent image) are amplified and silver clusters witha diameter of B0:6 mm are formed, which can beidentified by the optical pick-up of the scanningsystem, i.e. by a microscope equipped with a CCDcamera.

Automatic scanning systems have completelychanged our idea of nuclear emulsion. Nowadaysnuclear emulsion can be considered as a long termdata storage device which allows to recordcomplete three-dimensional information oncharged particle trajectory. This information isred out by automatic scanning systems.

3. Automatic scanning systems

The concept of full automatic track recognition,which is the base of the algoritm used in thecurrent systems, was proposed already in the 1970s[3,4] at the University of Nagoya (Japan). Un-fortunately, the technology was not advancedenough to implement it. The group continued theeffort to develop a full automatic system during1980s, and we have a first complete application ofthe automatic system in CHORUS experimentdata analysis in the 1990s. Nowadays the Nagoyagroup is close to deliver a third generation ofautomatic scanning system [5] and an intenseR&D program is underway also in Europe byseveral groups of the CHORUS Collaborations. InFig. 1 a schematic view of a typical automaticscanning system can be seen.

3.1. The Nagoya system

The basic components of the Track Selector(TS) are designed to detect tracks with predictedangle in the field of view of the CCD camera. Thesystem uses an Fast Programmable Gate Array(FPGA)+Fast Memory and a grabber boardconnected to a CCD Camera (512 � 512 pixel at120 Hz frame rate). The area of view is B150 �

Fig. 1. Schematic view of the components of a typical automatic scanning system for nuclear emulsion.

N. D’Ambrosio / Nuclear Instruments and Methods in Physics Research A 477 (2002) 431–434432

150 mm2: It is determined by the microscope opticsand by the surface of the CCD. The focal depth ofthe microscope gives an image relative to B3 mmof emulsion thickess. The track detection algo-rithm is simple. Sixteen tomographic images of(e.g.) 100 mm thick emulsion layers are taken anddigitised. Every digitised image is shifted horizon-tally relative to the first layer, so that the predictedangle tracks become perpendicular to the emulsionsurface. By superimposing, the 16 shifted digitisedimages tracks are identified as enhancements ofpulse heights. The quality of one track segment ismeasured by the signal pulse height. Outputs fromthe TS are the position ðx; y; zÞ and the angleðyx; yyÞ of the detected track segments. Currentlythe TS is at its second generation with New TrackSelector system. A third generation called ultratrack selector (UTS) is now operating and a Super-UTS is under development. The UTS system iscapable to analyse 1 cm2/h and recognise track upto 400 mrad with a track finding efficiency B98%:The Super-UTS is expected to increase thescanning speed up to 10 cm2/h. The scanningpower road map of the Nagoya systems has shownin Fig. 2.

3.2. Multi track system

Groups of the CHORUS Collaboration arefollowing the approach of the so-called Multi-Track Systems (MTS), initiated by the Salernogroup with theSySal system [6]. In the following

we present the R&D underway by the Naples andM .unster groups. Similar studies are being made bythe CERN and NIKHEF groups.

In an MTS system all tracks in each field of vieware reconstructed regardless of their slope, pro-vided they are within a region of angularacceptance. The raw data are series of tomo-graphic images of the emulsion (from 10 to 50),taken at different depth levels (2:523:5 mm). Adigital filter is used to clean each image. Athreshold is applied to the filter output to extractthe dark spots that are candidates to becomegrains. The following step consists in combininggrains from different layers to recognise geome-trical alignments. The field of view is subdivided incells about 20 mm wide. Local alignments of grainsare then detected within each cell and acrossboundaries of neighbouring cells. Track segmentsin the two emulsion layers are finally connectedand stored on the local database. The drivingprinciples in the design of this system is the use ofstate of the art commercial products, both for thehardware and for the software. This makes thesystem itself flexible enough to be upgradedfollowing the rapid progress in the technology.Preliminary tests were made using a largeplate microscope stage ð80 � 40 cm2Þ by Micos-gmbH.2 The microscope is equipped with faststepping motors or DC motors both on thehorizontal and on the vertical axes. Both a fastCCD camera ð512 � 512 pixel at 120 Hz) and high-resolution CCD camera (1024 � 1024 pixel at30 Hz) are used with this system. A field of viewof about 200 � 200 mm2 is provided with anintegration time of about 1ms. The performancecould be further improved by using fast MegaPixelcameras which will be soon available on themarket. They will allow an increase of the fieldof view up to 350 � 350 mm2 with, at least, 500 Hzframe rate.

The current system uses a Matrox Genesis3

board [7] with 1 Digital Signal Processor (DSP).This board is a commercial frame grabber andvideo images analysis device. The board isequipped with a Neighborhood Operation Accel-

Fig. 2. Scanning power road map of Nagoya system.

2Micos GMBH, 7224 Umkirch (Germany).3Genesis is a trademark of Matrox Electronic System Ltd.

N. D’Ambrosio / Nuclear Instruments and Methods in Physics Research A 477 (2002) 431–434 433

erator (NOA), a Matrox custom processor and adedicated DSP Texas Instruments TMS320C80,which contains four integer point processing unitsand a master processor. Up to six boards with 2DSP and 2 NOA can work together to analyse theimage. The system is capable of filtering, digitisingand clustering one frame in less than 10ms. Thisallows to envisage a speed of 1 mm/s during datataking from the emulsion. The time needed toanalyse a complete field (both emulsion sides)would thus be less then 300 ms. The tracking isperformed in parallel to the main acquisitionsequence. By using a Pentium II class processorwith 400 MHz clock, all tracks in a complete fieldare found in less than 200 ms. In perspective about3 fields per second could be analysed taking intoaccount the time needed to move to adjacentviews. About 14 min could then be required tofully reconstruct all tracks contained in 1 cm2

squared of ES.An approach oriented towards the use of

software and advanced commercial products inimage processing is also followed by the CERNand NIKHEF groups of CHORUS. They havedeveloped in collaboration with industry a newoptics with a large field of view and optimised itfor emulsion scanning. In the future, automaticscanning could profit from this development bothin speed and quality.

4. Automatic scanning systems for biophysics

The full programmability and flexibility ofcurrent systems for nuclear emulsions imageanalysis give the possibility to easily apply themin different research fields. In the Naples andM .unster laboratories, in collaboration with abiophysics group of Naples University, is understudy the feasibility of a full automatic system forchromosomal analysis, applying a technologytransfer between physics and biophysics. A semi-automatic system (operator-assisted) for chromo-somes counting is being tested at Naples Uni-versity. The speed of the system is 10 times fasterthan the manual scanning procedure. Chromo-

somal analysis is a routine method in severalmedical applications (prenatal diagnosis, cancercytogenetics, biological dosimetry). The mostimportant limiting factors in clinical cytogeneticsare the time of scoring, its accuracy and reprodu-cibility, and the sample size that can be analysed ina reasonable amount of time. The analysis ofchromosome images has many contact points withnuclear emulsion analysis and it can profit of theacquired experience and of the power of thecurrent systems.

5. Conclusions

The impressive progress of the automatic scan-ning systems technologies has stimulated therevival of nuclear emulsion as particle tracksdetector. The use of these systems together withnuclear emulsion allows the realisation of experi-ments with unprecedented statistics. On the otherhand the current systems are flexible enough to beused in any field where a fast digital image analysisis required.

Acknowledgements

I greatefully acknowledge the financial supportfrom EU (TMR program) under contract no.ERBFMRX-CT98-0196.

References

[1] E. Eskut, et al., (CHORUS Coll.), Nucl. Instr. and Meth. A

401 (1997) 7.

[2] CERN/SPSC 99-20 SPSC/P311 LNGS-LOI 19/99 August

5, 1999.

[3] S. Aoki, et al., Nucl. Instr. and Meth. B 51 (1990) 466.

[4] K. Niwa, et al., Proceedings of the International Cosmic

Ray Symposium on High Energy Phenomena, Cosmic Ray

Lab., Univ. Tokyo, 1974, p. 149.

[5] T. Nakano, Ph.D. Thesis, University of Nagoya, 1997.

[6] G. Rosa, et al., Nucl. Instr. and Meth. A 394 (1997) 357.

[7] Matrox Web page: http://www.matrox.com/imaging.

N. D’Ambrosio / Nuclear Instruments and Methods in Physics Research A 477 (2002) 431–434434