Detection of Watercore in Apples
Definition of Watercore
This paper discusses techniques for evaluating the amount of watercore present in apples and their applicability for packing line sorting. Research on this subject has been reported for over 30 years, with much of the early work being done in Wenatchee.
I would like to review here the nature and significance of watercore. A definition of watercore is:
"...a physiological disorder in apples that can cause the tissue of the apple to appear translucent and allow the intercellular spaces to be filled with a sugar-water solution." (Fidler et al., 1973)
Fidler's definition points to the underlying basis for detecting the presence of watercore-that of measuring a physical property of the apple that has been changed due to the presence of the "sugar-water solution in the intercellular spaces."
The problem with watercore is the internal damage that often develops in fruit affected with the disorder after a period of storage. According to Porritt et al. (1963), "Slight watercore often disappears during storage, but severe watercore, even if not detectable externally, can cause tissue breakdown if stored." This breakdown shows up as internal browning of the tissue.
Management of Watercore Apples
One of the challenges is to find a method, compatible with existing packing house operations, to detect watercore and sort fruit into separate categories so that storage and marketing decisions can be made with reliable knowledge of the condition of the fruit. While this is now done on a lot-by-lot basis using destructive assessments, packing house managers would like to have information about the extent of watercore present in each apple. They would like to have the ability to sort fruit individually according to the amount of watercore present in each apple.
The presence of the watery solution in the intercellular spaces leads to changes in certain physical properties and in the distribution of hydrogen atoms. The physical property changes that have been exploited by various detection devices include optical density, X-ray absorptivity, and mass density. The change in the spatial distribution of protons and the nature of their immediately surrounding environment has been detected by magnetic resonance imaging.
Light Transmittance Devices
Light transmittance devices are based on measurement of optical density (OD) differences. One early device was developed by Birth and Olsen (1964) in Wenatchee. First described in 1962, the device measured the difference in transmitted light intensity at two wavelengths-760 and 810 nanometers. The extent of watercore was expressed as a mathematical relationship:
Watercore index = OD760 - (0.8 x OD810)
Figure 1 shows a schematic representation of an optical density difference instrument. The device directed light of one chosen wavelength through the apple and into the photomultiplier detector. By shining various wavelengths of light through the apple and comparing the amount that was transmitted, a series of tests were performed that supported the watercore index shown above.
Figure 1. Schematic representation of optical density difference instrument.
How well did this method of detecting watercore work? In 1964, Birth and Olsen wrote about their tests in which 150 Delicious apples were studied. Those with medium degrees of watercore were excluded from the statistics. Using the watercore index shown above, they achieved a correlation coefficient, r, of -0.895. In a similar study, Francis et al. (1966) sampled Delicious apples twice monthly (100 each time, total 1,000). They achieved an average correlation coefficient, r, of -0.538. Despite the low correlation, internal browning was detected with 91% accuracy. Throop et al. (1994) used broad spectrum incandescent light and both color and monochrome digital cameras to produce a modernized version of the optical density difference instrument. They achieved 99% separation accuracy between unaffected and affected fruit. They also achieved 95% separation accuracy between slightly affected and moderately affected fruit. Best results were reported on fruit just after harvest. Their results excluded fruit with open calyxes and fruit that were misaligned during the reading.
In summary, researchers have reported poor to excellent correlation between optical density difference sorting laboratory results and observed watercore for a large group of apples. The best results have been achieved when attempting a separation between fruit that were unaffected from those that had any degree of watercore. Perhaps the most promising use of optical density sorting is for quality assurance checking where a small sample of apples can be manually inserted into a laboratory instrument one apple at a time. The difficulties that will have to be overcome prior to packing line implementation are the requirement to properly orient the fruit and the possible needs to block out all other light and to check for open calyxes.
Watery tissue absorbs more X-ray energy than does normal tissue, thus a device that is sensitive to changes in X-ray absorptivity has the potential to detect and sufficiently quantify the extent of watercore so that a separation could be made. The literature documents three methods for using X-ray to detect and sort watercore apples. The oldest is exposure of photographic film to the X-rays that have passed through an apple. These images are called radiographs. More recent electronic systems attempt to record the images digitally. Computed Tomography (CT) scanning creates images with multiple exposures to X-ray. This technique can produce very high quality quantitative measurements of internal characteristics. Line-Scan devices use one-dimensional linear detector arrays. Multiple line scans are needed to create one image. These devices are less expensive but produce relatively poorer internal quantitative information.
Recent results using X-ray have been reported in the literature. Schatzki et al. (1997) used both radiographs and line scan images. They were examining whether humans could view the images and detect watercore. Greater than 50% accuracy was obtained when oriented images were examined without a time limit. When time limits were imposed using digital line scan images on a computer screen, accuracy fell off rapidly as the image speed increased and the time for examination decreased. At approximately 50% of typical packing line speed, only 20% of watercore affected fruit (Red Delicious) were correctly identified. These results suggest that the images contain sufficient information to sort the fruit but that a computerized image analysis algorithm rather than human graders is necessary.
Another recent study reported in the literature was by Shahin and Tollner (1997). They used a line-scan system to examine methods for analyzing the digitized images to determine to what extent watercore was present. They examined 108 Red Delicious apples with 64%, on average, being correctly classified. Figure 2 shows a typical X-ray line scan on the left and the computer-enhanced zones of watercore on the right. Table 1 lists the computer-determined classification accuracy by degree of watercore.
Figure 2. X-ray line scans of Red Delicious apples. Raw images are on the left. Computer enhanced zones of watercore are depicted on the right (Shahin and Tollner, 1997)
Table 1. X-ray classification accuracy (from Shahin and Tollner, 1997).
|Actual severity class||
% Correctly classified
X-ray sorting systems may be adaptable from existing potato hollow heart packing line detection systems. Shahin and Tollner's work suggests that X-ray systems may be most accurate when separating fruit with no watercore from those that exhibit some watercore (75% correct) and least accurate when separating fruit with moderate degree of watercore from no watercore or severe watercore categories (40% correct).
Mass Density Sorting
Apples with different amounts of watercore have different mass densities (g/cm3). Hung et al. (1989) found that, in 1988, Washington Red Delicious apples had mass densities of 856, 864, 890, 925 and 940 kg/m3 with no, slight, moderate, heavy and severe watercore, respectively. In a study by Porritt et al. (1963), 136,000 bushels of Delicious apples were sorted in a flotation separator that used an alcohol-water mixture to create a fluid of reduced density. Seventy-eight percent were sorted correctly. A project at Washington State University (WSU) to design a two-phase fluid hydraulic sorting system capable of separating slightly affected and unaffected apples from those with greater amounts of watercore has been reported (Cavalieri et al., 1996). In that study, done in cooperation with Dr. Nate Reed of Stemilt Growers, Inc., watercore grading was done according to a six-level scale. Figure 3 depicts the range of mass density differences versus degree of watercore found in the WSU study.
Figure 3. Mass density of delicious apples as affected by degree of watercore (Cavalieri et al., 1996).
A sorting system that passes the apples through a region of low density fluid has the ability to cause the heavier (more dense) apples to sink while leaving the lighter (less dense) apples on the surface. Whereas Porritt et al. used a mixture of alcohol and water to create a fluid less dense than water, WSU's approach was to reduce the density by injection of a mist of air bubbles. Compressed air was introduced into a section of a conveying flume through which the apples were moving; 245 Red Delicious apples were tested with three replicates of each test. Five different sets of operating conditions were examined (changing separator plate height, etc.). This was a blind test in which the staff at WSU was unaware of the degree of watercore in the fruit. After sorting, the fruit were shipped to Stemilt Growers where they were destructively evaluated for degree of watercore. The results were that 100% of the apples with "moderate" or greater watercore (densities greater than 900 kg/m3) were correctly separated from those with less watercore. Eighty to ninety percent of those with only "slightly moderate" watercore (less than 900 kg/m3 but greater than 880 kg/m3) were separated from those with "slight" or "none" (densities less than 880 kg/m3).
Mass density sorting of watercore apples can be done very effectively (90-100% accuracy). The size of the apple was unimportant when mass density was 900 kg/m3 or greater. Thus, use prior to size sorting would be feasible. In the future, it may possible to use a combination of weight sizing and digital imaging equipment to electronically calculate the density of each apple. This would accomplish the same result as the flotation sorting systems. Throop and colleagues (1994) have explored this concept.
Magnetic Resonance Imaging
Referred to as MRI and based on the principles of Nuclear Magnetic Resonance (NMR), this technology detects the concentration of hydrogen nuclei in an object with respect to the tissue immediately surrounding them. The imaging tests done so far have used equipment virtually identical to medical imaging systems.
Clark and Bieleski (1997) from HortResearch in New Zealand have used MRI to follow the disappearance of watercore during storage in Fuji apples. At this time the cost and speed of MRI do not make it a viable contender for a packing line system. However, the quality of the information that it produces is excellent. Figures 4 and 5 show the disappearance of watercore in stored apples over time. Both "block" and "radial" type watercore are depicted. Clark and Bieleski state that for New Zealand Fuji, where watercore is not strongly correlated with fruit density, MRI may be the best system.
While MRI may not appear practical in the near future, Zion et al. (1997) have reported on the development of a non-imaging NMR system for detecting the presence of pits in olives for packing line application and report that the results to date are promising. Thus, while the analysis necessary to quantify the degree of watercore from an MR image exceeds the capabilities of today's electronics, the progress that has been demonstrated in using NMR to detect more sharply differing materials (pits versus edible tissue) is encouraging.
ConclusionsThere are several promising technologies being researched that should permit packing house operators to sort fruit affected with watercore. Mass density separation may be the least expensive when watercore is well related to fruit density. X-ray line-scan technology should be suitable for packing line use, but more work is needed to improve the analysis of the images to increase the separation accuracy under non-oriented packing line conditions. MRI and NMR technologies may be suitable for packing line use in the future, but great improvements in computational speed are needed. Optical density difference has been shown to be effective for quantifying the degree of watercore in individual fruit, but no packing line equipment has ever been developed.
Birth, G. S. and K. L. Olsen. 1964. Nondestructive detection of water core in Delicious apples. Proc. Amer. Soc. Hort. Sci. 85:74-84.
Cavalieri, R. P., G. M. Hyde and P. R. MacQuarrie. 1996. Hydraulic sorting of watercore apples. PH'96 International Postharvest Science Conference, Taupo, 4-9 August. pp. 78-86.
Clark, C. J. and R. L. Bieleski. 1997. Application of NMR imaging to detection of watercore in Fuji apple post-harvest. In: Proceedings from the Sensors for Nondestructive Testing International Conference and Tour. Orlando, 18-21 February. pp. 211-220.
Fidler, J. C., B. G. Wilkinson, K. L. Edney and R. O. Sharples. 1973. The Biology of Apple and Pear Storage. Commonwealth Agricultural Bureau, England. pp. 113-116.
Francis, F. J., W. J. Bramlage. and W. J. Lord. 1966. Detection of watercore and internal breakdown in Delicious apples by light transmittance. Proc. Amer. Soc. Hort. Sci. 87:78-84.
Hung, Y. C., E. W. Tollner and B. L. Upchurch. 1989. Physical properties of apples affected by watercore disorder. ASAE Paper No. 89-6507. ASAE, 2950 Niles Rd., St. Joseph, MI 49085-9659.
Porritt, S. W., A. D. McMechan and K. Williams. 1963. Note on a flotation method for segregation of water core apples. Can. J. Plant Sci. 43:600-602.
Schatzki, T. F., R. P. Haff, R. Young, I. Can, L. C. Le and N. Toyofuku. 1997. Defect detection in apples by means of X-ray imaging. In: Proceedings from the Sensors for Nondestructive Testing International Conference and Tour. Orlando, 18-21 February. pp. 161-171.
Shahin, M. A. and R. W. Tollner. 1997. Detection of watercore in apples using X-ray linescans feature extraction and classification. In: Proceedings from the Sensors for Nondestructive Testing International Conference and Tour. Orlando, 18-21 February. pp. 389-400.
Throop, J. A., D. J. Aneshansley and B. L. Upchurch. 1994. Camera system effects on detecting watercore in 'Red Delicious' apples. Trans. ASAE. 37:873-877.
Zion, B., S. M. Kim, M. J. McCarthy and P. Chen. 1997. On-line detection of pits in olives by NMR projections. In: Proceedings from the Sensors for Nondestructive Testing International Conference and Tour. Orlando, 18-21 February. pp. 376-386.
Biological Systems Engineering Department,
Washington State University
P.O. Box 646120, Pullman, WA 99164-6120
Fax: (509) 335-2722
Tree Fruit Postharvest Journal 8(4):3-8