Specimen Temperature Detection on a Clinical Laboratory Pre-Analytic Automation Track: Implications for Direct-from-Track Total Laboratory Automation (TLA) Systems

Abstract

Clinical laboratory regulations require temperature monitoring of facilities, reagent and specimen storage, as well as temperature-dependent equipment. Real-time specimen temperature detection has not yet been integrated into total laboratory automation (TLA) solutions. An infrared (IR) pyrometer was paired with a complementary metal oxide semiconductor (CMOS) laser sensor and connected to an embedded networked personal computer (PC) to create a modular temperature detection unit for closed, moving clinical laboratory specimens. Accuracy of the detector was assessed by comparing temperature measurements to those obtained from thermocouples connected to battery-operated data loggers. The temperature detector was then installed on a pre-analytic laboratory automation system to assess specimen temperature before and after processing on an integrated thawing and mixing (T/M) robotic workcell. The IR temperature detector was able to accurately record temperature of closed, moving specimens on a pre-analytic automation system. The effectiveness of the T/M workcell was independently verified using the temperature detector. Specimen reroute on the pre-analytic automation track was identified as a potential risk for frozen specimens being inadvertently delivered to future, connected instrumentation. Automated IR temperature detection can be used to verify specimen temperature prior to instrument loading and/or sampling. Such systems could be used to prevent frozen specimens from being inadvertently loaded onto analytical instrumentation in TLA solutions.

Introduction

Temperature monitoring of laboratory facilities and equipment such as refrigerators, freezers, water baths, and incubators is critical to ensuring that processes are conducted under appropriate conditions and in guaranteeing that specimens and reagents are stored appropriately.^1,2 Analytical instrumentation also frequently includes temperature-controlled mechanisms to stabilize reaction conditions or as a fundamental component of testing (e.g., PCR). Given the importance of temperature on laboratory processes, federal regulations under the Clinical Laboratory Improvement Amendments of 1988 (CLIA’88) require that laboratories define and monitor temperature conditions appropriate for storage of reagents and specimens, and in the operation of test systems, as applicable. ³ Temperature monitoring is therefore included as a requirement of CLIA-approved laboratory accreditation organizations such as the College of American Pathologists (CAP) Laboratory Accreditation Program, ⁴ and environmental conditions such as temperature are also addressed in standards included in ISO 15189. ⁵

For example, specimen storage temperature is specifically highlighted in requirements regarding appropriate handling of referred specimens (e.g., CAP checklist item GEN.40125). ⁴ Indeed, numerous pre-analytic processes involve moving specimens between different temperature conditions. In the reference laboratory setting, specimens are often received and/or maintained at different temperature conditions [e.g., room temperature (20–25 °C), refrigerated (2–8 °C), and/or frozen (less than −10 °C)] based on unique requirements of the testing that was ordered. Specimens are then delivered to the appropriate laboratory section(s) and, if frozen, thawed prior to analysis.

ARUP Laboratories has previously developed automated thawing and mixing (T/M) robotic workcells, which are directly connected to a customized pre-analytic automation system that receives specimens from our specimen receiving department.^6,7 These workcells are used to bring specimens to room temperature and then mix them thoroughly prior to delivery to laboratory section–specific sorters. Along with a goal for increased direct-from-track total laboratory automation (TLA), we recognized the need for a modular temperature detection unit that could assess the temperature of moving, closed-tube specimens on our automation track. The present article describes the development and testing of an infrared (IR) temperature detector designed to meet this goal. While online temperature monitoring is commonly used in both manufacturing and food processing, this is the first report of integration for this purpose into clinical laboratory pre-analytic automation. As such, it has relevance for both TLA solutions and potential integration into analytical instrumentation.

Materials and Methods

General

This project was administratively reviewed by the University of Utah Institutional Review Board, which determined that it did not meet federal definitions of human subject research.

Hardware and Software

The temperature detector processing unit ( Fig. 1A ) consists of a personal computer (PC) (CX9020; Beckhoff Automation, Balingen, Germany), an eight-channel digital input/output (I/O) module (EL1259; Beckhoff), an analog input module (EL3751; Beckhoff), and an external control push-button. The optical tube detection trigger (e.g., a tube detector; Fig. 1B, a ) consists of a complementary metal oxide semiconductor (CMOS) laser sensor (ZH500C3P; Keyence, Osaka, Japan) connected to the digital I/O module. The temperature detector ( Fig. 1B, b )—an IR pyrometer (thermoMETER LTF-CF1-C3; Micro-Epsilon, Ortenburg, Germany)—connects to an analog input in the processing unit. This temperature detector has a 9 ms response time, a temperature range from −50 °C to 975 °C, and a 3 mm focus at 150 mm distance from the lens. Embedded software in the processing unit is configured on a TwinCAT 3 control system (Beckhoff) conforming to open international standard IEC 61131-3. Additional programming runs the push-button, indicator logic to start and stop recording, indicators of network connectivity and system status (e.g., idle, recording, and errors), and indications of when values have been recorded and sent to the server. The computer unit and detectors are mounted, with the tube detector aimed before (e.g., up-track from) the subsequent temperature detector ( Fig. 1C ) and at an angle to maximize the time moving tubes are visible in the detector’s focal window (Fig. 1B and 1C).

Figure 1.

Infrared (IR) temperature detection system. (A) Data acquisition and control module. (B) Photograph of the installed detection system, consisting of the (a) tube detector and (b) IR temperature detector mounted to the MagneMotion automation track. The arrow shows the direction of specimen tube movement. (C) Diagram showing the operational strategy. Specimen tubes on the track move from right to left.

Event Triggering, Temperature Detection, and Data Recording

The process for triggering tube events and recording temperatures is as follows. When a moving tube on the automation track passes in front of the tube detector beam, this trigger event signals the program to wait 10 ms (for the temperature detector response time), then begin collecting data from the temperature detector through the following processes: (1) An analog 0–10 VDC (voltage direct current) signal is recorded at 10 kHz and placed in an array, (2) an analog input module runs an infinite impulse response filter on the data array at 1 kHz intervals and then passes those data to the main program running on the embedded PC, and (3) the main program collects all of these data into an array and then averages that for the final value when the tube detector sensor turns off.

A network application connects via Ethernet to the device’s embedded PC and monitors several variables, including connection integrity, failure logging, and user recording (e.g., start and stop) commands. When a user presses the device push-button, the application creates a new comma-separated value (CSV) file for data logging. When connected and running, the application watches for changes in a program “count” variable, retrieves the latest temperature measurement from the embedded PC, and enters a new record into the output CSV file. Each record in the CSV file contains the current date and time, the tube count value, and the temperature in degrees Celsius.

Verification of Temperature Detector: Nonconveyed Specimens

To verify that the temperature detector accurately measures temperature in liquid-filled aliquot tubes routinely used by our laboratory, the following experiment was conducted. Four ARUP standard transport tubes (polypropylene, cat no. 62.612; Sarstedt, Nümbrecht, Germany) were filled with demineralized distilled water (ddH₂O) and capped. Type K thermocouple probes (beaded thermocouple wire probe, −40 to 250 °C, cat no. P-01; REED Instruments, Wilmington, NC) were inserted through a small hole in the cap and positioned within the center of the fluid. Tubes were then frozen upright at −80 °C. Immediately prior to the experiment, the separate thermocouple wires were connected to a four-channel thermocouple thermometer/data logger (cat no. SD-947; REED Instruments) calibrated to a National Institute of Standards and Technology (NIST)-traceable thermometer. Tubes were then removed from dry ice and placed into nonmoving pucks on a circular research and development automation track (MagneMotion; Rockwell Automation, Devens, MA). Every 3 min, tubes were manually moved sequentially in front of the IR temperature detector, and the temperature for each tube was simultaneously recorded from the IR detector liquid crystal display (LCD) window and the REED thermometer. Temperature recordings were made every 3 min for 2 h.

Verification of Temperature Detector: Moving Specimens

To verify that the temperature detector accurately measures temperature from moving specimens, K-type thermocouple probes were frozen at <10 °C in ARUP standard transport tubes filled with ddH₂O. Immediately prior to experiments, a tube was removed from the freezer, and the thermocouple wire was connected to a Digi-Sense USB battery-operated temperature data logger (cat. no. DO-20250-44; Davis Instruments, Vernon Hills, IL) mounted to the back of a movable specimen puck on the research and development automation track. The data logger was positioned on the back of the puck so that it was out of the focal area of the IR detector. The data logger was configured to record temperature every 6 s. The puck, tube, and data logger were then circulated on a test track that can operate at variable speeds of 0.2 to 2.0 m/s, repeatedly past the mounted IR temperature detector with each loop. Data from the USB data logger were then imported into Excel 2016 (Microsoft, Redmond, WA) and aligned by time stamp to data acquired by the IR temperature detector. The thermocouple adjustment period (due to plugging the connector end of the frozen probe into an ambient data logger) was excluded from analysis.

Temperature Measurements from the Clinical Laboratory Automation Track

The temperature detector was then mounted on our clinical laboratory pre-analytic automation track at positions before (Fig. 2, position a) and after (Fig. 2, position b) T/M workcells, and the temperature of 5000 consecutive specimen tubes was measured at each position. The temperature detector was then mounted immediately prior to a clinical laboratory sorter (Fig. 2, position c), and the temperature of 45,000 consecutive specimen tubes was measured.

Figure 2.

Simplified diagram of the pre-analytic automation track and installation locations for the temperature detector. Specimens are loaded by the specimen receiving department onto the pre-analytic automation track (left). Specimens are then transported to thawing and mixing (T/M) robotic workcells if appropriate for ordered tests. Specimens are next transported to sorters designated for different laboratory sections, and subsorting into standardized racks occurs. Once testing is complete, specimens are loaded in standardized racks into storage sorters, which transfer tubes to both temperature-specific (room temperature, refrigerated, or frozen) and storage duration–specific racks for offline storage. If additional testing is pending on a specimen tube, it is rerouted back onto the track and delivered to its next assigned destination laboratory sorter. Positions marked (a), (b), and (c) indicate installation locations for the temperature detector (a) before and (b) after a T/M workcell, as well as (c) before a destination sorter. An actual automation track includes multiple T/M workcells, sorters, and storage sorters.

Analysis

Data analysis was conducted in Excel 2016 (Microsoft) and SigmaPlot 13 (Systat, San Jose, CA). Graphs were generated using SigmaPlot and Publisher 2016 (Microsoft). Statistical significance was assessed using the Wilcoxon signed-rank test in SigmaPlot. P-values < 0.05 were considered significant.

Results

Figure 1 shows the temperature detector’s control unit ( Fig. 1A ), detector configuration when installed on the automated track ( Fig. 1B ), and focal design in relation to moving specimens ( Fig. 1C ). Preliminary experiments established that adding a background material (e.g., ~1 ft. ² fixed panel) behind the tube detection window increased the signal-to-noise ratio in data acquisition. A background panel was therefore included in subsequent experiments of temperature detection with moving specimens.

A comparison of temperature measurements from the IR temperature detector versus thermocouples frozen in ddH₂O-filled standard transport tubes as they thawed in nonconveyed pucks on the research and development test track is shown in Figure 3A–3C . IR temperature detector measurements were consistently higher than those from the embedded thermocouple (p < 0.01); see Figure 3B . These differences were greater at frozen temperatures (detector temperature differences of >4 °C in specimens that were <10 °C as measured by the thermocouple). Temperature differences were intermediate in specimens that were thawing or at immediate postthaw temperatures (~3 °C detector temperature difference at specimen temperatures from 0 to 6 °C), while temperature differences were smaller as the specimens approached room temperature (<1.5 °C detector temperature difference at specimen temperatures >12 °C). Minor calibration adjustments were therefore made in the IR temperature detection software to account for these differences in subsequent studies, because the above data suggest that the outside of specimen tubes may be slightly warmer than the frozen and/or thawing fluid inside.

Figure 3.

Temperature measurements from the infrared (IR) temperature detector versus thermocouples in nonconveyed containers. (A) Corresponding temperature measurements (°C) from the IR temperature detector (○) and thermocouples (●) frozen in demineralized distilled water (ddH₂O) thawing throughout time in nonconveyed specimen tubes (n = 4; data shown as mean ± SD). (B) Difference in mean temperature (IR temperature detector measurement − thermocouple measurement, in °C) throughout time in the same specimen tubes as shown in (A). (C) Temperature measurements from IR temperature detector (y-axis) plotted against temperature measurements from thermocouples (x-axis) in the same specimen tubes as shown in (A). Dotted line, linear regression: y = 0.91x + 3.23, r² = 0.999.

The IR temperature detection system was then tested on moving tubes on the research and development automation track. Figure 4 shows a representative example of a ddH₂O-filled tube, previously frozen with an embedded thermocouple attached to a battery-powered data logger, comparing temperature measurements from the IR temperature detector versus those from the thermocouple. Temperature measurements from both methods tracked closely, although increased variability in measurements was observed from the IR temperature detection system versus the thermocouple. For example, the coefficient of variation (CV%) for temperature measurements during the first 10 min, as displayed in Figure 4 (a relative temperature plateau vs. other time points), was 103.1% for the IR temperature detector versus 14.5% for the thermocouple.

Figure 4.

Temperature measurements from the infrared (IR) temperature detector versus a thermocouple in a moving container. (A) Corresponding temperature measurements (°C) from a moving frozen container [filled with demineralized distilled water (ddH₂O)] using the IR temperature detector (○) and a thermocouple frozen in the specimen tube (●) connected to a battery-operated USB data logger. Recordings were made during the time while the specimen tube was transported on a moving puck on the research and development automation track loop.

To assess how the IR temperature detection system would perform in a clinical laboratory setting, the device was then installed in three different locations on our clinical laboratory pre-analytic track (Fig. 2, positions a, b, and c). As background, ARUP implemented a customized MagneMover LITE (Rockwell Automation) automation system in 2015. ⁷ In this system, specimens received from our specimen receiving department (frozen, refrigerated, and ambient) are loaded via pick-and-place binder robots from a belt-driven track system onto a high-speed magnetic track, which moves specimens in pucks at speeds of up to 2.0 m/s. Depending on the unique requirements associated with the laboratory assays ordered, tubes are then directed automatically to T/M workcells and/or specimen sorters.

Figure 5 shows consecutive temperature measurements of tubes entering ( Fig. 5A ; n = 5000) and leaving ( Fig. 5B ; n = 5000) the automated T/M workcells. Because only one temperature detector has been developed, these represent different datasets acquired from separate installation locations (positions noted in Fig. 2 ). While the preponderance of specimens on the track are at room temperature, the temperature detector was able to detect many colder (refrigerated and/or frozen) specimens ( Fig. 5A ), as would be expected of tubes coming from the specimen receiving department. The effectiveness of the T/M robotic stations in bringing specimens to room temperature was also apparent ( Fig. 5B ).

Figure 5.

Temperature measurements from the infrared (IR) temperature detector placed before and after a thaw–mix (T/M) workcell. (A) Five thousand consecutive temperature measurements from clinical specimens entering a T/M workcell. (B) Five thousand separate consecutive temperature measurements from clinical specimens exiting a T/M workcell. (C) Temperature measurements from 45,000 consecutive specimens entering a rack-based robotic sorter. See Figure 2 for detector installation positions on the track. Dashed line is 0 °C for reference.

The temperature detector was then installed near the input of a final sorter (i.e., a rack-based destination for tubes for one of the automated laboratories; see Fig. 2, position c). Figure 5C shows consecutive temperature measurements of 45,000 tubes directed to this sorter. Because this laboratory is operational 24 h a day, the track is programmed such that specimens destined for testing in this laboratory predominantly go through T/M workcells prior to sorting. A small number of frozen and refrigerated specimens, however, also passed to this sorter ( Fig. 5C ). On further investigation, it was found that these events represented specimen reroutes, in which testing had previously been performed by another laboratory section, the specimens had been stored frozen (or refrigerated) to preserve sample stability per assay protocol, and the track was then used to sort specimens for subsequent testing of additional components (see Fig. 2 , reroute pathway). The temperature detector was able to identify these tubes, and may therefore be useful for providing a safeguard to prevent delivery of frozen specimens to instrumentation in future direct-from-track TLA configurations.

Discussion

This article describes the development of a closed-tube IR temperature detection system that was installed and tested on a custom clinical laboratory pre-analytic automated track. While specimen integrity checks (e.g., hemolysis, lipemia, and icterus), liquid level detection, and even phlebotomy tube type detection (e.g., cap color recognition) have been incorporated into analytical and/or TLA systems, we are unaware of any other implementation of specimen temperature detection in a clinical laboratory automation track. This article demonstrated the risk of inadvertently loading frozen specimens onto an automation system, and proposes a mechanism to detect (and ultimately divert) such specimens in future TLA solutions.

Loading of a frozen specimen on an automated direct-from-track TLA system (or onto an instrument directly) could lead to instrument probe crashes and/or inaccurate specimen sampling. This may affect analysis from not just the frozen specimen, but also subsequent specimens if a probe is not functioning properly and the malfunction is not quickly identified. Temperature detection systems could mitigate this risk, ideally being used to divert frozen specimens to appropriate robotics (e.g., T/M workcells) or holding lanes if such robotics are unavailable. In analytical instrumentation, automated detection of frozen specimens could be used to prevent subsequent probe crashes.

IR temperature detection methods are limited by inherent background noise associated with data acquisition. While installation of a flat panel (e.g., cardboard) behind the moving specimens helped to reduce background noise, initial experiments demonstrated that very short acquisition times combined with quickly moving specimens could lead to erroneously high temperature measurements. Installation of the temperature detector on slower segments of the track (e.g., input/output lanes; see Fig. 2 ) resolved these erroneous measurements and allowed for the accurate identification of frozen specimens. Another possible installation strategy would be to incorporate IR temperature detection with barcode-scanning modules, because these processes typically involve stationary (or near-stationary) tubes. Given the critical importance of accurate barcode scanning in clinical laboratory specimen identification, we did not attempt this configuration in the present investigation.

The present studies also demonstrated that there could be differences between the external temperature of a specimen tube and the internal specimen temperature. While these were usually relatively small (see Figs. 2 and 3), the present experiments did not determine whether differences may be greater in tubes with decreased specimen volumes. We also did not investigate whether different numbers or types of specimen labels can affect accurate temperature detection using IR methods. In addition, laboratories that incorporate biobanked specimens maintained at colder temperatures (e.g., −80 °C or −200 °C) may benefit from temperature detection systems that can accurately measure temperatures lower than the range of the IR pyrometer used in this article (−50 °C to 975 °C).

While precise measurement of specimen temperature may have limited value in a routine clinical laboratory setting (where freezing of specimens is not a part of pre-analytic processes), send-out testing to referral laboratories frequently requires frozen specimens to preserve sample stability during transportation. Because more reference laboratories are implementing direct-from-track TLA solutions to improve turnaround time, this may inadvertently confer an increased risk of frozen specimens being loaded onto automated analytical systems. IR temperature detection systems could mitigate this risk and provide specimen-specific temperature monitoring with far greater resolution than current systems, which focus primarily on the environment or temperature-specific equipment but not the specimens themselves.